@vocalstack/mcp 1.3.0 → 1.5.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -61,21 +61,21 @@ Both offer the same tools.
61
61
  ## Tools
62
62
 
63
63
  <!-- tools -->
64
- - `transcribe`: Start transcribing a finished recording, from a public link or a file sent inline; returns a transcriptionId at once, before any text.
65
- - `get_transcription`: Read one transcription's status and text, with timestamps and speakers: as it was spoken or translated, at once or after waiting for it to finish.
64
+ - `create_transcription`: Start transcribing a finished recording, from a public link or a file sent inline; returns a transcriptionId at once, before any text.
65
+ - `get_transcription`: Read one transcription's status and text, with timestamps and speakers, or its title and summary: as spoken or translated, at once or after waiting for it to finish.
66
66
  - `list_transcriptions`: List this account's transcriptions, newest first, a page at a time: files, links and live ones, including those broadcasts recorded.
67
67
  - `search_transcriptions`: Find what was said across this account's transcriptions, or in one: the lines with given words or a name (mode words), or the passages that answer a question (mode meaning).
68
- - `get_summary`: Get the title and summary of a finished transcription, writing one the first time if it has none.
69
- - `rename_speaker`: Name a speaker in one transcription, so its transcript shows the name instead of "Speaker N".
68
+ - `update_transcription`: Change a transcription: its name, its public share link, its summary, or whether spoken translations use the speakers' own cloned voices.
70
69
  - `delete_transcription`: Move a transcription to the trash, with its recordings, transcript, translations and summary; its share links stop working.
71
- - `restore_transcription`: Bring a deleted transcription back from the trash, with its recordings, transcript, translations, summary and working share links.
72
- - `set_cloned_voices`: Turn cloned voices on or off for one transcription's spoken translations.
73
70
  - `list_speakers`: List the speakers this account has saved: people recognised by name in every new transcription.
74
71
  - `save_speaker`: Save one speaker of a transcription to the account's saved speakers, so later transcriptions recognise and name them; their own speech in this transcription is the sample.
72
+ - `update_speaker`: Change a speaker: name one person in one transcription (instead of "Speaker N"), or rename a saved speaker and switch their cloned voice on or off, everywhere they are recognised.
75
73
  - `delete_speaker`: Delete a saved speaker for good: their name, fingerprint, stored sample and any cloned voice are erased at once and cannot be restored.
76
74
  - `list_broadcasts`: List this account's broadcasts, or get one in detail.
77
75
  - `start_broadcast`: Start live transcription and translation of a stream playing now, on a public link listeners open in their own language.
76
+ - `update_broadcast`: Change a broadcast's settings: its name, who can listen (public, or private with a password), the language spoken, and whether translations are read aloud, in stock or the speakers' own cloned voices.
78
77
  - `stop_broadcast`: Stop the live transcription a broadcast is running, which also stops its billing.
78
+ - `delete_broadcast`: Move a broadcast to the trash: its public link and stream addresses stop working for listeners.
79
79
  <!-- /tools -->
80
80
 
81
81
  Errors come back as tool results with `isError` set and a body with `error` (a code such as `not_found`, `invalid_options`, `invalid_state`, `insufficient_credit` or `rate_limited`), `message`, `retryable`, `retryAfterS` and `topUpUrl` when credit is needed. Retry a start with the same `idempotencyKey` so nothing is made or billed twice.
package/dist/handlers.js CHANGED
@@ -301,7 +301,7 @@ function decodeFile(a) {
301
301
  throw new ToolError("invalid_audio", "the file is empty");
302
302
  return bytes;
303
303
  }
304
- async function transcribe(a, ctx) {
304
+ async function createTranscription(a, ctx) {
305
305
  const key = idempotencyKey(a);
306
306
  const file = sourceOf(a);
307
307
  const options = transcribeOptions(a);
@@ -472,7 +472,7 @@ async function translatedLines(ctx, id, snap, t, lang, offset, limit) {
472
472
  async function getTranscription(a, ctx) {
473
473
  const id = a.id("transcriptionId");
474
474
  const waitS = a.int("waitS", 0, 0, WAIT_MAX_S);
475
- const format = a.oneOf("format", ["text", "segments", "none"], "text");
475
+ const format = a.oneOf("format", ["text", "segments", "summary", "none"], "text");
476
476
  const lang = await targetLang(a, ctx, "lang");
477
477
  const words = a.bool("includeWords");
478
478
  if (words && lang !== undefined)
@@ -488,11 +488,25 @@ async function getTranscription(a, ctx) {
488
488
  t = await transcription(ctx, id);
489
489
  }
490
490
  }
491
- const out = { status: statusOf(t) };
491
+ const status = statusOf(t);
492
+ try {
493
+ const links = await ctx.api.get(path("/v1/transcriptions/", id, "/share-links"), undefined, ctx.signal);
494
+ const first = links?.links?.[0];
495
+ if (first?.url)
496
+ status.shareUrl = first.url;
497
+ }
498
+ catch { }
499
+ const out = { status };
492
500
  if (timedOut !== undefined)
493
501
  out.timedOut = timedOut;
494
502
  if (format === "none")
495
503
  return out;
504
+ if (format === "summary") {
505
+ if (words)
506
+ throw bad("`includeWords` goes with format segments");
507
+ out.summary = await summaryOf(ctx, id, t, lang);
508
+ return out;
509
+ }
496
510
  if (lang !== undefined) {
497
511
  let notYet;
498
512
  if (t.job && t.job.state !== "done")
@@ -616,11 +630,7 @@ async function searchMeaning(a, ctx) {
616
630
  async function search(a, ctx) {
617
631
  return a.oneOf("mode", ["words", "meaning"], "words") === "meaning" ? searchMeaning(a, ctx) : searchWords(a, ctx);
618
632
  }
619
- async function summary(a, ctx) {
620
- const id = a.id("transcriptionId");
621
- const lang = await targetLang(a, ctx, "lang");
622
- const t = await transcription(ctx, id);
623
- const pending = (message) => ({ state: "pending", message });
633
+ async function summaryOf(ctx, id, t, lang) {
624
634
  const s = await ctx.api.get(path("/v1/transcriptions/", id, "/summary"), { lang }, ctx.signal);
625
635
  if (s.state === "ready") {
626
636
  const v = { state: "ready", title: s.title ?? t.name, summary: s.summary };
@@ -631,20 +641,28 @@ async function summary(a, ctx) {
631
641
  return { state: "unavailable", message: "The transcription failed, so there is nothing to summarise." };
632
642
  }
633
643
  if ((t.job && !TERMINAL.has(t.job.state)) || (!t.job && OPEN.has(t.state))) {
634
- return pending("The transcription is still running; wait for it (get_transcription with waitS), then ask again.");
644
+ return { state: "pending", message: "The transcription is still running; wait for it (get_transcription with waitS), then ask again." };
635
645
  }
636
646
  if (s.state === "pending")
637
- return pending("The summary is being written; ask again in a few seconds.");
638
- let made;
639
- try {
640
- made = (await ctx.api.request("POST", path("/v1/transcriptions/", id, "/summarize"), { signal: ctx.signal })).data;
647
+ return { state: "pending", message: "The summary is being written; ask again in a few seconds." };
648
+ if (s.state === "unavailable")
649
+ return { state: "unavailable", message: "Summaries are not available on this server." };
650
+ const length = t.durationS ?? t.job?.source?.durationS;
651
+ if (typeof length === "number" && length < 60) {
652
+ return { state: "not_requested", message: "Transcriptions shorter than a minute are not summarised." };
641
653
  }
642
- catch (e) {
643
- if (e instanceof ToolError && e.code === "forbidden") {
644
- throw new ToolError("forbidden", "This transcription has no summary yet, and starting one needs the write scope; this credential can only read an existing summary.", { requiredScope: ["write"] });
645
- }
646
- throw e;
654
+ return { state: "none", message: "No summary was asked for; update_transcription with summary true writes one." };
655
+ }
656
+ async function writeSummary(ctx, id, t) {
657
+ const current = await summaryOf(ctx, id, t);
658
+ if (current.state === "ready" || current.state === "unavailable" || current.state === "not_requested")
659
+ return current;
660
+ if ((t.job && !TERMINAL.has(t.job.state)) || (!t.job && OPEN.has(t.state))) {
661
+ throw new ToolError("invalid_state", "The transcription is still running; wait for it (get_transcription with waitS), then ask for the summary.");
647
662
  }
663
+ if (current.state === "pending")
664
+ return current;
665
+ const made = (await ctx.api.request("POST", path("/v1/transcriptions/", id, "/summarize"), { signal: ctx.signal })).data;
648
666
  switch (made.state) {
649
667
  case "ready":
650
668
  return { state: "ready", title: made.title ?? t.name, summary: made.summary };
@@ -653,7 +671,7 @@ async function summary(a, ctx) {
653
671
  case "not_requested":
654
672
  return { state: "not_requested", message: "Transcriptions shorter than a minute are not summarised." };
655
673
  default:
656
- return pending("The summary is being written; ask again in a few seconds.");
674
+ return { state: "pending", message: "The summary is being written; ask again in a few seconds." };
657
675
  }
658
676
  }
659
677
  async function all(ctx, route, field, query = {}) {
@@ -804,16 +822,126 @@ async function deleteTranscription(a, ctx) {
804
822
  put(out, "restorableUntil", t.purgeAfter);
805
823
  return out;
806
824
  }
807
- async function restoreTranscription(a, ctx) {
808
- const id = a.id("transcriptionId");
809
- await transcription(ctx, id);
810
- const t = (await ctx.api.request("POST", path("/v1/transcriptions/", id, "/restore"), { signal: ctx.signal })).data;
811
- return { transcriptionId: t.id, name: t.name, restored: true };
825
+ function optBool(a, k) {
826
+ const v = a.raw[k];
827
+ if (v === undefined || v === null)
828
+ return undefined;
829
+ if (typeof v !== "boolean")
830
+ throw bad(`\`${k}\` must be true or false`);
831
+ return v;
832
+ }
833
+ function speakerTarget(a) {
834
+ const sid = present(a, "speakerId");
835
+ const tid = present(a, "transcriptionId");
836
+ const spk = present(a, "speaker");
837
+ if (sid && !tid && !spk)
838
+ return true;
839
+ if (!sid && tid && spk)
840
+ return false;
841
+ if (sid)
842
+ throw bad("give either `speakerId` (a saved speaker) or `transcriptionId` with `speaker`, not both");
843
+ if (tid)
844
+ throw bad("`speaker` is required with `transcriptionId`");
845
+ if (spk)
846
+ throw bad("`transcriptionId` is required with `speaker`");
847
+ throw bad("give `transcriptionId` with `speaker` (one transcription) or `speakerId` (a saved speaker)");
848
+ }
849
+ function noSavedSpeaker(e) {
850
+ if (e instanceof ToolError && (e.code === "not_found" || (e.code === "forbidden" && !e.details.requiredScope))) {
851
+ throw new ToolError("not_found", NO_SPEAKER);
852
+ }
853
+ throw e;
854
+ }
855
+ async function savedSpeaker(ctx, id) {
856
+ try {
857
+ return await ctx.api.get(path("/v1/speakers/", id), undefined, ctx.signal);
858
+ }
859
+ catch (e) {
860
+ noSavedSpeaker(e);
861
+ }
862
+ }
863
+ async function patchSpeaker(ctx, id, body) {
864
+ try {
865
+ await ctx.api.request("PATCH", path("/v1/speakers/", id), { json: body, signal: ctx.signal });
866
+ }
867
+ catch (e) {
868
+ noSavedSpeaker(e);
869
+ }
870
+ }
871
+ async function updateSavedSpeaker(a, ctx) {
872
+ const id = a.id("speakerId");
873
+ let name;
874
+ const given = a.raw.name;
875
+ if (given !== undefined && given !== null) {
876
+ if (typeof given !== "string")
877
+ throw bad("`name` must be a string");
878
+ name = given.trim();
879
+ const n = [...name].length;
880
+ if (n === 0 || n > 60)
881
+ throw bad("a saved speaker's `name` must be 1 to 60 characters");
882
+ }
883
+ const cloned = optBool(a, "clonedVoice");
884
+ const consent = a.bool("consent");
885
+ if (name === undefined && cloned === undefined)
886
+ throw bad("give `name`, `clonedVoice` or both to change a saved speaker");
887
+ if (cloned === true && !consent) {
888
+ throw bad("Turning a saved speaker's cloned voice on needs consent: true, confirming that person agreed to have their voice cloned.");
889
+ }
890
+ const report = (s) => {
891
+ const out = {
892
+ speakerId: s.id,
893
+ name: s.name,
894
+ consent: s.consent,
895
+ clonedVoice: s.voice?.enabled === true && s.consent === "granted",
896
+ };
897
+ if (s.voice)
898
+ put(out, "voiceState", s.voice.state);
899
+ return out;
900
+ };
901
+ if (cloned === undefined) {
902
+ try {
903
+ return report((await ctx.api.request("PATCH", path("/v1/speakers/", id), { json: { name }, signal: ctx.signal })).data);
904
+ }
905
+ catch (e) {
906
+ noSavedSpeaker(e);
907
+ }
908
+ }
909
+ const cur = await savedSpeaker(ctx, id);
910
+ if (cloned === true && cur.consent === "refused") {
911
+ throw new ToolError("invalid_state", "This speaker refused to be cloned and their audio was deleted; save them again from a transcription (save_speaker with consent) to clone them.");
912
+ }
913
+ if (name !== undefined)
914
+ await patchSpeaker(ctx, id, { name });
915
+ if (cloned === true) {
916
+ let s;
917
+ try {
918
+ s = (await ctx.api.request("POST", path("/v1/speakers/", id, "/consent"), { json: { consent: "granted" }, signal: ctx.signal })).data;
919
+ }
920
+ catch (e) {
921
+ if (e instanceof ToolError && e.code === "invalid_state" && e.message.includes("no sample stored")) {
922
+ throw new ToolError("invalid_state", "This saved speaker has no stored sample to clone their voice from; save them again from a transcription (save_speaker with consent).");
923
+ }
924
+ noSavedSpeaker(e);
925
+ }
926
+ if (s?.voice && s.voice.enabled === false)
927
+ await patchSpeaker(ctx, id, { enabled: true });
928
+ }
929
+ else if (cloned === false && cur.voice?.enabled === true) {
930
+ await patchSpeaker(ctx, id, { enabled: false });
931
+ }
932
+ return report(await savedSpeaker(ctx, id));
812
933
  }
813
- async function renameSpeaker(a, ctx) {
934
+ async function updateSpeaker(a, ctx) {
935
+ if (speakerTarget(a))
936
+ return updateSavedSpeaker(a, ctx);
937
+ if (present(a, "clonedVoice")) {
938
+ throw bad("`clonedVoice` is for a saved speaker (speakerId); for everyone in one transcription use update_transcription with clonedVoices");
939
+ }
814
940
  const id = a.id("transcriptionId");
815
941
  const number = a.int("speaker", 1, 1, 1000);
816
942
  const given = a.raw.name;
943
+ if (given === undefined || given === null)
944
+ throw bad('`name` is required with `transcriptionId` and `speaker` ("" clears the name)');
817
945
  if (typeof given !== "string")
818
946
  throw bad('`name` must be a string ("" clears the name)');
819
947
  const name = [...given.trim()].slice(0, 80).join("");
@@ -833,27 +961,119 @@ async function renameSpeaker(a, ctx) {
833
961
  return { transcriptionId: id, speaker: number, name: name || null };
834
962
  }
835
963
  const VOICE_CLONING_NEEDS_CONSENT = "turning voiceCloning on needs voiceCloningConsent: true — confirm you have permission from everyone speaking";
836
- async function setClonedVoices(a, ctx) {
964
+ const NEEDS_CONSENT = "Turning cloned voices on needs consent: true, confirming everyone speaking agreed to have their voice cloned.";
965
+ function consentError(e) {
966
+ if (e instanceof ToolError && e.message === VOICE_CLONING_NEEDS_CONSENT)
967
+ throw bad(NEEDS_CONSENT);
968
+ throw e;
969
+ }
970
+ async function updateTranscription(a, ctx) {
837
971
  const id = a.id("transcriptionId");
838
- const on = a.raw.on;
839
- if (typeof on !== "boolean")
840
- throw bad("`on` must be true or false");
972
+ let name = a.str("name");
973
+ if (name !== undefined) {
974
+ name = name.trim();
975
+ const n = [...name].length;
976
+ if (n === 0 || n > 200)
977
+ throw bad("`name` must be 1 to 200 characters");
978
+ }
979
+ const share = optBool(a, "shareLink");
980
+ const summary = optBool(a, "summary");
981
+ const cloned = optBool(a, "clonedVoices");
841
982
  const consent = a.bool("consent");
842
- await transcription(ctx, id);
843
- let t;
983
+ if (summary === false)
984
+ throw bad("`summary` takes only true (write one now); a summary is never deleted");
985
+ if (name === undefined && share === undefined && summary === undefined && cloned === undefined) {
986
+ throw bad("give at least one of `name`, `shareLink`, `summary` or `clonedVoices` to change");
987
+ }
988
+ if (cloned === true && !consent)
989
+ throw bad(NEEDS_CONSENT);
990
+ let t = await transcription(ctx, id);
991
+ if (t.lifecycle && t.lifecycle !== "active") {
992
+ throw new ToolError("invalid_state", "This transcription is in the trash; restore it in the VocalStack Dashboard first.");
993
+ }
994
+ const written = summary === true ? await writeSummary(ctx, id, t) : undefined;
995
+ if (name !== undefined || cloned !== undefined) {
996
+ const body = {};
997
+ if (name !== undefined)
998
+ body.name = name;
999
+ if (cloned !== undefined)
1000
+ body.settings = { voiceCloning: cloned, voiceCloningConsent: consent };
1001
+ try {
1002
+ t = (await ctx.api.request("PATCH", path("/v1/transcriptions/", id), { json: body, signal: ctx.signal })).data;
1003
+ }
1004
+ catch (e) {
1005
+ consentError(e);
1006
+ }
1007
+ }
1008
+ let links = (await ctx.api.get(path("/v1/transcriptions/", id, "/share-links"), undefined, ctx.signal))?.links ?? [];
1009
+ if (share === true && links.length === 0) {
1010
+ const l = (await ctx.api.request("POST", path("/v1/transcriptions/", id, "/share-links"), { json: { label: "Shared link" }, signal: ctx.signal })).data;
1011
+ links = [l];
1012
+ }
1013
+ else if (share === false) {
1014
+ for (const l of links)
1015
+ await ctx.api.request("DELETE", path("/v1/share-links/", l.id), { signal: ctx.signal });
1016
+ links = [];
1017
+ }
1018
+ const out = { transcriptionId: t.id, name: t.name, shareUrl: links[0]?.url ?? null, clonedVoices: t.settings?.voiceCloning === true };
1019
+ if (written)
1020
+ out.summary = written;
1021
+ return out;
1022
+ }
1023
+ async function updateBroadcast(a, ctx) {
1024
+ const id = a.id("broadcastId");
1025
+ const name = a.str("name");
1026
+ const access = a.raw.access === undefined || a.raw.access === null ? undefined : a.oneOf("access", ["public", "private"], "public");
1027
+ const password = a.str("password");
1028
+ const spokenRaw = a.str("spokenLanguage");
1029
+ const speak = optBool(a, "spokenTranslations");
1030
+ const cloned = optBool(a, "clonedVoices");
1031
+ const consent = a.bool("consent");
1032
+ if ([name, access, password, spokenRaw, speak, cloned].every((v) => v === undefined)) {
1033
+ throw bad("give at least one of `name`, `access`, `password`, `spokenLanguage`, `spokenTranslations` or `clonedVoices` to change");
1034
+ }
1035
+ if (cloned === true && !consent)
1036
+ throw bad(NEEDS_CONSENT);
1037
+ await broadcast(ctx, id);
1038
+ const body = {};
1039
+ if (name !== undefined)
1040
+ body.name = name;
1041
+ if (access !== undefined)
1042
+ body.access = access;
1043
+ if (password !== undefined && access !== "public")
1044
+ body.password = password;
1045
+ const settings = {};
1046
+ if (spokenRaw !== undefined) {
1047
+ const l = spokenRaw.trim().toLowerCase();
1048
+ settings.spokenLanguage = l === "" || l === "auto" ? null : l;
1049
+ }
1050
+ if (speak !== undefined)
1051
+ settings.translationTts = speak;
1052
+ if (cloned !== undefined) {
1053
+ settings.voiceCloning = cloned;
1054
+ settings.voiceCloningConsent = consent;
1055
+ }
1056
+ if (Object.keys(settings).length)
1057
+ body.settings = settings;
1058
+ let b;
844
1059
  try {
845
- t = (await ctx.api.request("PATCH", path("/v1/transcriptions/", id), {
846
- json: { settings: { voiceCloning: on, voiceCloningConsent: consent } },
847
- signal: ctx.signal,
848
- })).data;
1060
+ b = (await ctx.api.request("PATCH", path("/v1/broadcasts/", id), { json: body, signal: ctx.signal })).data;
849
1061
  }
850
1062
  catch (e) {
851
- if (e instanceof ToolError && e.message === VOICE_CLONING_NEEDS_CONSENT) {
852
- throw bad("Turning cloned voices on needs consent: true, confirming everyone speaking agreed to have their voice cloned.");
1063
+ if (e instanceof ToolError && e.code === "invalid_state") {
1064
+ throw new ToolError("invalid_state", "This broadcast is in the trash; restore it in the VocalStack Dashboard first.");
853
1065
  }
854
- throw e;
1066
+ if (e instanceof ToolError && e.message === "private broadcasts need a password") {
1067
+ throw bad("A private broadcast needs a `password` (at least 4 characters) the first time.");
1068
+ }
1069
+ consentError(e);
855
1070
  }
856
- return { transcriptionId: t.id, name: t.name, clonedVoices: t.settings?.voiceCloning === true };
1071
+ return {
1072
+ ...brief(b),
1073
+ spokenLanguage: b.settings?.spokenLanguage ?? "auto",
1074
+ spokenTranslations: b.settings?.translationTts === true,
1075
+ clonedVoices: b.settings?.voiceCloning === true,
1076
+ };
857
1077
  }
858
1078
  const ON_AIR = "This broadcast is already on air; stop it first (stop_broadcast) or use another broadcast.";
859
1079
  async function startBroadcast(a, ctx) {
@@ -945,20 +1165,38 @@ async function stopBroadcast(a, ctx) {
945
1165
  const after = await transcription(ctx, tid);
946
1166
  return { ...liveStatus(after), broadcastId: b.id, link: b.shareUrl };
947
1167
  }
1168
+ async function deleteBroadcast(a, ctx) {
1169
+ const id = a.id("broadcastId");
1170
+ await broadcast(ctx, id);
1171
+ let b;
1172
+ try {
1173
+ b = (await ctx.api.request("POST", path("/v1/broadcasts/", id, "/archive"), { signal: ctx.signal })).data;
1174
+ }
1175
+ catch (e) {
1176
+ if (e instanceof ToolError && e.code === "invalid_state" && e.message.includes("stop the recording first")) {
1177
+ throw new ToolError("invalid_state", "This broadcast is on air; stop it first (stop_broadcast), then delete it.", { broadcastId: id });
1178
+ }
1179
+ throw e;
1180
+ }
1181
+ const out = { broadcastId: b.id, name: b.name, deleted: true };
1182
+ put(out, "deletedAt", b.archivedAt);
1183
+ put(out, "restorableUntil", b.purgeAfter);
1184
+ return out;
1185
+ }
948
1186
  export const handlers = {
949
- transcribe,
1187
+ create_transcription: createTranscription,
950
1188
  get_transcription: getTranscription,
951
1189
  list_transcriptions: listTranscriptions,
952
1190
  search_transcriptions: search,
953
- get_summary: summary,
954
- rename_speaker: renameSpeaker,
1191
+ update_transcription: updateTranscription,
955
1192
  delete_transcription: deleteTranscription,
956
- restore_transcription: restoreTranscription,
957
- set_cloned_voices: setClonedVoices,
958
1193
  list_speakers: speakers,
959
1194
  save_speaker: saveSpeaker,
1195
+ update_speaker: updateSpeaker,
960
1196
  delete_speaker: deleteSpeaker,
961
1197
  list_broadcasts: listBroadcasts,
962
1198
  start_broadcast: startBroadcast,
1199
+ update_broadcast: updateBroadcast,
963
1200
  stop_broadcast: stopBroadcast,
1201
+ delete_broadcast: deleteBroadcast,
964
1202
  };
package/dist/tools.json CHANGED
@@ -1,9 +1,9 @@
1
1
  {
2
- "instructions": "VocalStack transcribes and translates recorded or live audio and video in about 100 languages, with speaker labels.\nRecorded audio: transcribe (a link or a file) returns a transcriptionId at once; get_transcription with waitS waits for it and reads the text, in any language with lang.\nFind: list_transcriptions (ids, newest first); search_transcriptions (mode words for words or names, meaning for a question).\nOne transcription: get_summary; rename_speaker; set_cloned_voices; delete_transcription and restore_transcription.\nLive streams: start_broadcast, then get_transcription while it runs, stop_broadcast to end it; list_broadcasts shows them.\nPeople: list_speakers, save_speaker (voice cloned only with the person's consent), delete_speaker.\nOnly transcription is billed; everything else, and listening, is free. Retry a start with the same idempotencyKey.\nErrors are results with isError true: error, message, retryable, and retryAfterS or topUpUrl when they apply.",
2
+ "instructions": "VocalStack transcribes and translates recorded or live audio and video in about 100 languages, with speaker labels.\nRecorded audio: create_transcription (a link or a file) returns a transcriptionId at once; get_transcription with waitS waits for it and reads the text, in any language with lang.\nFind: list_transcriptions (ids, newest first); search_transcriptions (mode words for words or names, meaning for a question).\nOne transcription: get_transcription with format summary for its title and summary; update_transcription (name, share link, summary, cloned voices); delete_transcription (to the trash, restorable in the Dashboard).\nLive streams: start_broadcast, then get_transcription while it runs, stop_broadcast to end it; list_broadcasts shows them; update_broadcast changes access, language and spoken translations; delete_broadcast removes one.\nPeople: list_speakers, save_speaker (voice cloned only with the person's consent), update_speaker (names, and a saved speaker's cloned voice), delete_speaker.\nOnly transcription is billed; everything else, and listening, is free. Retry a start with the same idempotencyKey.\nErrors are results with isError true: error, message, retryable, and retryAfterS or topUpUrl when they apply.",
3
3
  "tools": [
4
4
  {
5
- "name": "transcribe",
6
- "title": "Transcribe a recording",
5
+ "name": "create_transcription",
6
+ "title": "Create a transcription",
7
7
  "description": "Start transcribing a finished recording, from a public link or a file sent inline; returns a transcriptionId at once, before any text.\nUse: audio or video that has ended (a file, podcast episode, YouTube video or public post). A stream playing now: start_broadcast. The result: get_transcription with waitS, no polling needed.\nList price: $0.20 an hour of audio, billed per second transcribed; Pay as you go bills at least 5 minutes per transcription. Each call is a new, billed transcription unless it repeats an idempotencyKey; a refused call costs nothing; summary and shareLink are free.\nNew transcriptions an hour per account: 10 on Pay as you go, 100 on Premium.\nTranscriptions running at once per account: 5 on Pay as you go, 50 on Premium; more wait their turn instead of failing.\nReturns: transcriptionId, state (queued, fetching, transcribing, then done, failed or stopped), progress, shareUrl when shareLink is true, replayed true for a repeated key.\nErrors: invalid_options (neither or both of url and fileBase64, not a public http(s) link, bad base64, too large), invalid_audio (empty file), idempotency_key_reused, insufficient_credit (with topUpUrl), quota_exceeded, rate_limited (with retryAfterS). A source that cannot be read ends in state failed, with error.",
8
8
  "inputSchema": {
9
9
  "additionalProperties": false,
@@ -84,7 +84,7 @@
84
84
  },
85
85
  "summary": {
86
86
  "default": false,
87
- "description": "Also write a title and summary when it is done, at no extra cost (get_summary reads it). Never for recordings under a minute.",
87
+ "description": "Also write a title and summary when it is done, at no extra cost (get_transcription with format summary reads it). Never for recordings under a minute.",
88
88
  "type": "boolean"
89
89
  },
90
90
  "url": {
@@ -101,7 +101,7 @@
101
101
  "type": "object"
102
102
  },
103
103
  "annotations": {
104
- "title": "Transcribe a recording",
104
+ "title": "Create a transcription",
105
105
  "readOnlyHint": false,
106
106
  "destructiveHint": false,
107
107
  "idempotentHint": false,
@@ -111,16 +111,17 @@
111
111
  {
112
112
  "name": "get_transcription",
113
113
  "title": "Read a transcription",
114
- "description": "Read one transcription's status and text, with timestamps and speakers: as it was spoken or translated, at once or after waiting for it to finish.\nUse: after transcribe (waitS saves polling), or any time during a live transcription for the text so far. An overview: get_summary. Finding a passage: search_transcriptions.\nBehaviour: free. A wait ends as soon as the work does, and sends progress notifications when the client gives a progress token. lang needs a finished transcription; the first call in a new language can take up to a minute on a long one, later pages are quick. Page with offset = nextOffset.\nReturns: status (state, progress, error when failed), timedOut when waitS was set, then segmentCount, partial while text is still coming, text or segments, nextOffset when more remain, and lang and translated with lang. A wait that runs out with lang set returns the status only.\nErrors: not_found; invalid_state for lang before it is finished; invalid_options for an unsupported lang or includeWords with lang; rate_limited when waitS is set and 4 waits are already open for this key (8 for the account).",
114
+ "description": "Read one transcription's status and text, with timestamps and speakers, or its title and summary: as spoken or translated, at once or after waiting for it to finish.\nUse: after create_transcription (waitS saves polling), or any time during a live transcription for the text so far. An overview: format summary. Finding a passage in many transcriptions: search_transcriptions.\nBehaviour: free. A wait ends as soon as the work does, and sends progress notifications when the client gives a progress token. lang needs a finished transcription; the first call in a new language can take up to a minute on a long one, later pages are quick. Page with offset = nextOffset.\nReturns: status (state, progress, error when failed, shareUrl when it has a public link: made with shareLink in create_transcription or update_transcription), timedOut when waitS was set, then segmentCount, partial while text is still coming, text or segments, nextOffset when more remain, and lang and translated with lang. format summary returns summary instead: state ready with title and summary (and lang when translated), pending while it is written or the transcription runs, none when none was asked for (update_transcription with summary true writes one), not_requested under a minute, unavailable when it failed. A wait that runs out with lang set returns the status only.\nErrors: not_found; invalid_state for lang before it is finished; invalid_options for an unsupported lang or includeWords with lang; rate_limited when waitS is set and 4 waits are already open for this key (8 for the account).",
115
115
  "inputSchema": {
116
116
  "additionalProperties": false,
117
117
  "properties": {
118
118
  "format": {
119
119
  "default": "text",
120
- "description": "text: one string, a line per segment as \"[mm:ss] Speaker 1: words\", for reading and quoting; segments: timed lines (start, end, text, speakerNumber, speakerName), for subtitles and exact times; none: status only (with waitS: wait without reading).",
120
+ "description": "text: one string, a line per segment as \"[mm:ss] Speaker 1: words\", for reading and quoting; segments: timed lines (start, end, text, speakerNumber, speakerName), for subtitles and exact times; summary: the title and summary only, for an overview; none: status only (with waitS: wait without reading).",
121
121
  "enum": [
122
122
  "text",
123
123
  "segments",
124
+ "summary",
124
125
  "none"
125
126
  ]
126
127
  },
@@ -130,7 +131,7 @@
130
131
  "type": "boolean"
131
132
  },
132
133
  "lang": {
133
- "description": "Code of a language to read the whole text in, one of about 100 (e.g. es, de, ja): the transcription must be finished (state done, or a stopped live one). The language already spoken returns the original with translated false. Omit for the original.",
134
+ "description": "Code of a language to read the whole text (or, with format summary, the summary) in, one of about 100 (e.g. es, de, ja): the transcription must be finished (state done, or a stopped live one). The language already spoken returns the original with translated false. Omit for the original.",
134
135
  "examples": [
135
136
  "es",
136
137
  "ja"
@@ -157,7 +158,7 @@
157
158
  "type": "integer"
158
159
  },
159
160
  "transcriptionId": {
160
- "description": "A transcriptionId from transcribe, list_transcriptions, search_transcriptions or start_broadcast.",
161
+ "description": "A transcriptionId from create_transcription, list_transcriptions, search_transcriptions or start_broadcast.",
161
162
  "examples": [
162
163
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
163
164
  ],
@@ -191,7 +192,7 @@
191
192
  {
192
193
  "name": "list_transcriptions",
193
194
  "title": "List transcriptions",
194
- "description": "List this account's transcriptions, newest first, a page at a time: files, links and live ones, including those broadcasts recorded.\nUse: to find a transcriptionId or see what exists. What was said: search_transcriptions. Broadcasts: list_broadcasts.\nBehaviour: free. lifecycle archived lists the trash, which restore_transcription can empty back for 30 days. A cursor this tool did not give out is refused.\nReturns: transcriptions (transcriptionId, name, createdAt, kind file or live, state, durationS and title when known, broadcastId for a broadcast's), and nextCursor while more remain. None: an empty list.\nErrors: invalid_options for an unknown cursor.",
195
+ "description": "List this account's transcriptions, newest first, a page at a time: files, links and live ones, including those broadcasts recorded.\nUse: to find a transcriptionId or see what exists. What was said: search_transcriptions. Broadcasts: list_broadcasts.\nBehaviour: free. lifecycle archived lists the trash: deleted transcriptions, which the VocalStack Dashboard can restore for 30 days. A cursor this tool did not give out is refused.\nReturns: transcriptions (transcriptionId, name, createdAt, kind file or live, state, durationS and title when known, broadcastId for a broadcast's), and nextCursor while more remain. None: an empty list.\nErrors: invalid_options for an unknown cursor.",
195
196
  "inputSchema": {
196
197
  "additionalProperties": false,
197
198
  "properties": {
@@ -232,7 +233,7 @@
232
233
  {
233
234
  "name": "search_transcriptions",
234
235
  "title": "Search transcriptions",
235
- "description": "Find what was said across this account's transcriptions, or in one: the lines with given words or a name (mode words), or the passages that answer a question (mode meaning).\nUse: words when you know what was said; meaning for a question or topic, in any language. Text around a hit: get_transcription with offset. One transcription's overview: get_summary.\nBehaviour: free; no paging. words ranks whole words first, then close spellings, then newest, and searches the query's English translation too. meaning returns up to 10 passages, closest first. Searches by meaning an hour per account: 60 on Pay as you go, no limit on Premium.\nReturns: words: query, queryEn, hits (transcriptionId, name, mediaStart in seconds, speaker, text, score, live, recordedAt, rendered with lang). meaning: query, state (ready, or unavailable or off when it cannot run), passages (transcriptionId, name, mediaStart, mediaEnd, text, similarity 0 to 1, live), searchesLeftThisHour when capped. Cite the transcription and second. No match: an empty list.\nErrors: not_found for an unknown transcriptionId; invalid_options for lang or limit with mode meaning; rate_limited (with retryAfterS) past the hourly searches by meaning.",
236
+ "description": "Find what was said across this account's transcriptions, or in one: the lines with given words or a name (mode words), or the passages that answer a question (mode meaning).\nUse: words when you know what was said; meaning for a question or topic, in any language. Text around a hit: get_transcription with offset. One transcription's overview: get_transcription with format summary.\nBehaviour: free; no paging. words ranks whole words first, then close spellings, then newest, and searches the query's English translation too. meaning returns up to 10 passages, closest first. Searches by meaning an hour per account: 60 on Pay as you go, no limit on Premium.\nReturns: words: query, queryEn, hits (transcriptionId, name, mediaStart in seconds, speaker, text, score, live, recordedAt, rendered with lang). meaning: query, state (ready, or unavailable or off when it cannot run), passages (transcriptionId, name, mediaStart, mediaEnd, text, similarity 0 to 1, live), searchesLeftThisHour when capped. Cite the transcription and second. No match: an empty list.\nErrors: not_found for an unknown transcriptionId; invalid_options for lang or limit with mode meaning; rate_limited (with retryAfterS) past the hourly searches by meaning.",
236
237
  "inputSchema": {
237
238
  "additionalProperties": false,
238
239
  "properties": {
@@ -273,7 +274,7 @@
273
274
  "type": "string"
274
275
  },
275
276
  "transcriptionId": {
276
- "description": "Search only this transcription (a transcriptionId from list_transcriptions or transcribe). Omit to search the whole account.",
277
+ "description": "Search only this transcription (a transcriptionId from list_transcriptions or create_transcription). Omit to search the whole account.",
277
278
  "examples": [
278
279
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
279
280
  ],
@@ -295,66 +296,40 @@
295
296
  }
296
297
  },
297
298
  {
298
- "name": "get_summary",
299
- "title": "Get a summary",
300
- "description": "Get the title and summary of a finished transcription, writing one the first time if it has none.\nUse: for an overview. Full text, or the whole text in another language: get_transcription. A specific question: search_transcriptions with mode meaning.\nBehaviour: free. Writing a first summary needs the write scope and keeps it for later calls; reading an existing one needs read only.\nReturns: state ready with title and summary (and lang when translated); pending while it is written or the transcription still runs (call again in a few seconds); not_requested for recordings under a minute; unavailable when the transcription failed.\nErrors: not_found; forbidden when a summary must be written and the key is read-only.",
299
+ "name": "update_transcription",
300
+ "title": "Update a transcription",
301
+ "description": "Change a transcription: its name, its public share link, its summary, or whether spoken translations use the speakers' own cloned voices.\nUse: renaming, sharing or unsharing, asking for a summary later, or switching cloned voices. Only the fields given change; give at least one. Speaker names: update_speaker. Removing it: delete_transcription.\nArguments: transcriptionId (required). name 1 to 200 characters. shareLink true makes a public read-only link (or keeps the one it has), false turns every public link off. summary true writes a title and summary if it has none (a finished transcription of at least a minute). clonedVoices true needs consent true, recording that everyone speaking agreed to be cloned; false needs no consent.\nBehaviour: free; calling again with the same fields changes nothing. A link turned off cannot be turned back on: shareLink true afterwards makes a new address. With cloned voices on, each speaker's translated words are read in that speaker's own voice, never anyone else's; speakers who refused, or without a usable clone, keep a stock voice; lines read after the change use it.\nReturns: transcriptionId, name, shareUrl (null with no public link), summary (state ready, pending, not_requested or unavailable) when asked for, clonedVoices.\nErrors: invalid_options for no field to change, a bad name, summary false, or clonedVoices true without consent; invalid_state for a deleted transcription, or summary before it is finished; not_found; forbidden for a read-only key.",
301
302
  "inputSchema": {
302
303
  "additionalProperties": false,
303
304
  "properties": {
304
- "lang": {
305
- "description": "Optional language code for the title and summary, honoured when the transcription was made with summary translations on; otherwise the original comes back without lang.",
306
- "examples": [
307
- "fr"
308
- ],
309
- "type": "string"
305
+ "clonedVoices": {
306
+ "description": "true: translations of each speaker's own words are read aloud in that speaker's own cloned voice (needs consent true); false: stock voices.",
307
+ "type": "boolean"
308
+ },
309
+ "consent": {
310
+ "default": false,
311
+ "description": "Must be true with clonedVoices true: confirms everyone speaking in it agreed to have their voice cloned. Recorded with your account and the time. Not needed to turn cloned voices off.",
312
+ "type": "boolean"
310
313
  },
311
- "transcriptionId": {
312
- "description": "A transcriptionId from transcribe, list_transcriptions, search_transcriptions or start_broadcast.",
313
- "examples": [
314
- "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
315
- ],
316
- "format": "uuid",
317
- "type": "string"
318
- }
319
- },
320
- "required": [
321
- "transcriptionId"
322
- ],
323
- "type": "object"
324
- },
325
- "annotations": {
326
- "title": "Get a summary",
327
- "readOnlyHint": false,
328
- "destructiveHint": false,
329
- "idempotentHint": true,
330
- "openWorldHint": false
331
- }
332
- },
333
- {
334
- "name": "rename_speaker",
335
- "title": "Rename a speaker",
336
- "description": "Name a speaker in one transcription, so its transcript shows the name instead of \"Speaker N\".\nUse: labelling who spoke here. To recognise the person in later transcriptions too: save_speaker (it also names them here). Saved speakers: list_speakers.\nBehaviour: free; applies to that person in every recording of the transcription, and listeners following it live see it at once.\nReturns: transcriptionId, speaker, name (null when cleared).\nErrors: not_found for an unknown transcription or speaker number; forbidden for a read-only key.",
337
- "inputSchema": {
338
- "additionalProperties": false,
339
- "properties": {
340
314
  "name": {
341
- "description": "The name to show instead of \"Speaker N\". An empty string clears it back to \"Speaker N\".",
315
+ "description": "New display name, shown in lists and on the share page; trimmed, 1 to 200 characters.",
342
316
  "examples": [
343
- "Maria Lopez"
317
+ "Board meeting, 19 November"
344
318
  ],
345
- "maxLength": 80,
319
+ "maxLength": 200,
320
+ "minLength": 1,
346
321
  "type": "string"
347
322
  },
348
- "speaker": {
349
- "description": "The N of \"Speaker N\" in get_transcription's text, or speakerNumber in its segments; it stays the same after naming.",
350
- "examples": [
351
- 2
352
- ],
353
- "minimum": 1,
354
- "type": "integer"
323
+ "shareLink": {
324
+ "description": "true: make a public link anyone can open to read the transcript without signing in (an existing one is kept; shareUrl in the result). false: every public link of it stops working at once.",
325
+ "type": "boolean"
326
+ },
327
+ "summary": {
328
+ "description": "true: write a title and summary now if it has none (free; the transcription must be finished and at least a minute long). Read it with get_transcription format summary. false is not accepted: a summary is never deleted.",
329
+ "type": "boolean"
355
330
  },
356
331
  "transcriptionId": {
357
- "description": "A transcriptionId from transcribe, list_transcriptions, search_transcriptions or start_broadcast.",
332
+ "description": "A transcriptionId from create_transcription, list_transcriptions, search_transcriptions or start_broadcast.",
358
333
  "examples": [
359
334
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
360
335
  ],
@@ -363,14 +338,12 @@
363
338
  }
364
339
  },
365
340
  "required": [
366
- "transcriptionId",
367
- "speaker",
368
- "name"
341
+ "transcriptionId"
369
342
  ],
370
343
  "type": "object"
371
344
  },
372
345
  "annotations": {
373
- "title": "Rename a speaker",
346
+ "title": "Update a transcription",
374
347
  "readOnlyHint": false,
375
348
  "destructiveHint": false,
376
349
  "idempotentHint": true,
@@ -380,12 +353,12 @@
380
353
  {
381
354
  "name": "delete_transcription",
382
355
  "title": "Delete a transcription",
383
- "description": "Move a transcription to the trash, with its recordings, transcript, translations and summary; its share links stop working.\nUse: to remove one. restore_transcription brings it back unchanged for 30 days; after that it is erased for good. Saved speakers stay (delete_speaker).\nBehaviour: a file or link still being transcribed is stopped first and keeps its text so far, billed only for what was transcribed; restoring does not resume it. A live one must be stopped first (stop_broadcast).\nReturns: transcriptionId, name, deleted, deletedAt, restorableUntil.\nErrors: invalid_state while recording live or when locked; not_found; forbidden for a read-only key.",
356
+ "description": "Move a transcription to the trash, with its recordings, transcript, translations and summary; its share links stop working.\nUse: to remove one. For 30 days it can be restored unchanged from the trash in the VocalStack Dashboard (list_transcriptions with lifecycle archived lists it); after that it is erased for good. Saved speakers stay (delete_speaker). A broadcast itself: delete_broadcast.\nBehaviour: a file or link still being transcribed is stopped first and keeps its text so far, billed only for what was transcribed; restoring does not resume it. A live one must be stopped first (stop_broadcast).\nReturns: transcriptionId, name, deleted, deletedAt, restorableUntil.\nErrors: invalid_state while recording live or when locked; not_found; forbidden for a read-only key.",
384
357
  "inputSchema": {
385
358
  "additionalProperties": false,
386
359
  "properties": {
387
360
  "transcriptionId": {
388
- "description": "A transcriptionId from list_transcriptions (lifecycle archived lists the trash) or transcribe.",
361
+ "description": "A transcriptionId from list_transcriptions or create_transcription.",
389
362
  "examples": [
390
363
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
391
364
  ],
@@ -407,52 +380,53 @@
407
380
  }
408
381
  },
409
382
  {
410
- "name": "restore_transcription",
411
- "title": "Restore a transcription",
412
- "description": "Bring a deleted transcription back from the trash, with its recordings, transcript, translations, summary and working share links.\nUse: undoing delete_transcription within 30 days; list_transcriptions with lifecycle archived lists the trash.\nBehaviour: free. A transcription stopped by its deletion stays stopped (transcribe it again to finish it).\nReturns: transcriptionId, name, restored true.\nErrors: not_found for an unknown id or one erased after its 30 days; forbidden for a read-only key.",
383
+ "name": "list_speakers",
384
+ "title": "List saved speakers",
385
+ "description": "List the speakers this account has saved: people recognised by name in every new transcription.\nUse: to see who is known or find a speakerId for delete_speaker. Save one: save_speaker. Names inside one transcription come from get_transcription and are set with update_speaker, which saves no one.\nBehaviour: free; no paging.\nReturns: speakers, newest first, each with id, name, gender when known, cloned (their voice was cloned, with consent, to read translations of their own words) and, when cloned, voiceReady. None saved: an empty list.",
413
386
  "inputSchema": {
414
387
  "additionalProperties": false,
415
- "properties": {
416
- "transcriptionId": {
417
- "description": "A transcriptionId from list_transcriptions (lifecycle archived lists the trash) or transcribe.",
418
- "examples": [
419
- "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
420
- ],
421
- "format": "uuid",
422
- "type": "string"
423
- }
424
- },
425
- "required": [
426
- "transcriptionId"
427
- ],
388
+ "properties": {},
428
389
  "type": "object"
429
390
  },
430
391
  "annotations": {
431
- "title": "Restore a transcription",
432
- "readOnlyHint": false,
392
+ "title": "List saved speakers",
393
+ "readOnlyHint": true,
433
394
  "destructiveHint": false,
434
395
  "idempotentHint": true,
435
396
  "openWorldHint": false
436
397
  }
437
398
  },
438
399
  {
439
- "name": "set_cloned_voices",
440
- "title": "Turn cloned voices on or off",
441
- "description": "Turn cloned voices on or off for one transcription's spoken translations.\nOn: each speaker's translated words are read in that speaker's own cloned voice, never anyone else's; a speaker without a usable clone gets a stock voice. Off: stock voices for all. Voices are chosen by VocalStack; this is the only voice choice.\nConsent: turning on records, with your account and the time, that everyone speaking agreed, and releases the clones of this transcription waiting for it. Speakers who refused are never cloned.\nBehaviour: free; lines read aloud after the change use the new setting.\nReturns: transcriptionId, name, clonedVoices.\nErrors: invalid_options for on without consent, invalid_state for a deleted transcription, not_found, forbidden for a read-only key.",
400
+ "name": "save_speaker",
401
+ "title": "Save a speaker",
402
+ "description": "Save one speaker of a transcription to the account's saved speakers, so later transcriptions recognise and name them; their own speech in this transcription is the sample.\nUse: remembering a person across transcriptions. This transcription only: update_speaker. Who is saved: list_speakers. Remove: delete_speaker.\nArguments: transcriptionId and speaker point at one person: speaker is the number in \"Speaker 2\" from get_transcription, who must have spoken (a live transcription can still be running). name is 1 to 60 characters, trimmed; renaming later: update_speaker with speakerId. consent is true only when that person agreed to be cloned; leaving it out or false keeps only their fingerprint.\nBehaviour: free; also names them in this transcription. Each call saves a new speaker, so check list_speakers first. Without consent only a voice fingerprint is kept; with it the voice is cloned too, recorded with your account and the time.\nReturns: speakerId, name, consent (granted or pending), cloned, voiceState (training, then ready) when cloned.\nErrors: not_found for an unknown transcription or speaker number; invalid_options for too little speech of them (the message says how much there is); forbidden for a read-only key.",
442
403
  "inputSchema": {
443
404
  "additionalProperties": false,
444
405
  "properties": {
445
406
  "consent": {
446
407
  "default": false,
447
- "description": "Must be true to turn cloned voices on: confirms everyone speaking agreed to have their voice cloned. Recorded with your account and the time. Not needed to turn them off.",
408
+ "description": "true only when this person agreed to have their voice cloned: it is then cloned too, from their own speech in this transcription, so translations of their words can be read in their voice. false: only their fingerprint is saved. Never set it without the person's agreement.",
448
409
  "type": "boolean"
449
410
  },
450
- "on": {
451
- "description": "true: read each speaker's translated words in their own cloned voice; false (every transcription's default): stock voices.",
452
- "type": "boolean"
411
+ "name": {
412
+ "description": "The person's name, shown in this and later transcriptions.",
413
+ "examples": [
414
+ "Maria Lopez"
415
+ ],
416
+ "maxLength": 60,
417
+ "minLength": 1,
418
+ "type": "string"
419
+ },
420
+ "speaker": {
421
+ "description": "The N of \"Speaker N\" in get_transcription's text, or speakerNumber in its segments. Works while a live transcription is still running.",
422
+ "examples": [
423
+ 2
424
+ ],
425
+ "minimum": 1,
426
+ "type": "integer"
453
427
  },
454
428
  "transcriptionId": {
455
- "description": "A transcriptionId from transcribe, list_transcriptions, search_transcriptions or start_broadcast.",
429
+ "description": "A transcriptionId from create_transcription, list_transcriptions, search_transcriptions or start_broadcast.",
456
430
  "examples": [
457
431
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
458
432
  ],
@@ -462,66 +436,62 @@
462
436
  },
463
437
  "required": [
464
438
  "transcriptionId",
465
- "on"
439
+ "speaker",
440
+ "name"
466
441
  ],
467
442
  "type": "object"
468
443
  },
469
444
  "annotations": {
470
- "title": "Turn cloned voices on or off",
445
+ "title": "Save a speaker",
471
446
  "readOnlyHint": false,
472
447
  "destructiveHint": false,
473
- "idempotentHint": true,
474
- "openWorldHint": false
475
- }
476
- },
477
- {
478
- "name": "list_speakers",
479
- "title": "List saved speakers",
480
- "description": "List the speakers this account has saved: people recognised by name in every new transcription.\nUse: to see who is known or find a speakerId for delete_speaker. Save one: save_speaker. Names inside one transcription come from get_transcription and are set with rename_speaker, which saves no one.\nBehaviour: free; no paging.\nReturns: speakers, newest first, each with id, name, gender when known, cloned (their voice was cloned, with consent, to read translations of their own words) and, when cloned, voiceReady. None saved: an empty list.",
481
- "inputSchema": {
482
- "additionalProperties": false,
483
- "properties": {},
484
- "type": "object"
485
- },
486
- "annotations": {
487
- "title": "List saved speakers",
488
- "readOnlyHint": true,
489
- "destructiveHint": false,
490
- "idempotentHint": true,
448
+ "idempotentHint": false,
491
449
  "openWorldHint": false
492
450
  }
493
451
  },
494
452
  {
495
- "name": "save_speaker",
496
- "title": "Save a speaker",
497
- "description": "Save one speaker of a transcription to the account's saved speakers, so later transcriptions recognise and name them; their own speech in this transcription is the sample.\nUse: remembering a person across transcriptions. This transcription only: rename_speaker. Who is saved: list_speakers. Remove: delete_speaker.\nBehaviour: free; also names them in this transcription. Each call saves a new speaker, so check list_speakers first. Without consent only a voice fingerprint is kept; with it the voice is cloned too, recorded with your account and the time.\nReturns: speakerId, name, consent (granted or pending), cloned, voiceState (training, then ready) when cloned.\nErrors: not_found for an unknown transcription or speaker number; invalid_options for too little speech of them (the message says how much there is); forbidden for a read-only key.",
453
+ "name": "update_speaker",
454
+ "title": "Update a speaker",
455
+ "description": "Change a speaker: name one person in one transcription (instead of \"Speaker N\"), or rename a saved speaker and switch their cloned voice on or off, everywhere they are recognised.\nUse: labelling who spoke here, correcting a saved speaker's name, or turning their cloned voice on with their consent. To recognise a person in later transcriptions: save_speaker (it also names them here). Saved speakers: list_speakers.\nArguments: give exactly one of transcriptionId with speaker (the number in \"Speaker 2\" from get_transcription, not a name or an id), with name; or speakerId (from list_speakers) with name, clonedVoice or both. A name in a transcription is cut to 80 characters and an empty one undoes it; a saved speaker's is 1 to 60. clonedVoice true needs consent true.\nBehaviour: free and reversible; calling again replaces the name. In a transcription it applies to that person in every recording of it, and listeners following live see it at once; their speaker number never changes. clonedVoice true clones a saved speaker from their stored sample if they have no clone yet (voiceState training, then ready); false keeps the clone, unused (delete_speaker erases it). Nothing else about the person changes.\nReturns: transcriptionId, speaker, name (null when cleared); or, with speakerId, speakerId, name, consent, clonedVoice, and voiceState when cloned.\nErrors: invalid_options for both or neither kind of target, nothing to change, a bad name, clonedVoice with transcriptionId, or clonedVoice true without consent; invalid_state for a saved speaker who refused cloning or has no sample; not_found for an unknown transcription, speaker number or speakerId; forbidden for a read-only key.",
498
456
  "inputSchema": {
499
457
  "additionalProperties": false,
500
458
  "properties": {
459
+ "clonedVoice": {
460
+ "description": "With speakerId only: true reads translations of this person's own words in their own cloned voice (needs consent true; cloned from their stored sample if not yet), false back to a stock voice (the clone is kept, unused).",
461
+ "type": "boolean"
462
+ },
501
463
  "consent": {
502
464
  "default": false,
503
- "description": "true only when this person agreed to have their voice cloned: it is then cloned too, from their own speech in this transcription, so translations of their words can be read in their voice. false: only their fingerprint is saved. Never set it without the person's agreement.",
465
+ "description": "Must be true with clonedVoices true: confirms this person agreed to have their voice cloned. Recorded with your account and the time. Not needed to turn cloned voices off.",
504
466
  "type": "boolean"
505
467
  },
506
468
  "name": {
507
- "description": "The person's name, shown in this and later transcriptions.",
469
+ "description": "The name to show instead of \"Speaker N\", trimmed: up to 80 characters in one transcription (an empty string clears it back to \"Speaker N\"), 1 to 60 for a saved speaker. Required with transcriptionId.",
508
470
  "examples": [
509
- "Maria Lopez"
471
+ "Maria Lopez",
472
+ ""
510
473
  ],
511
- "maxLength": 60,
512
- "minLength": 1,
474
+ "maxLength": 80,
513
475
  "type": "string"
514
476
  },
515
477
  "speaker": {
516
- "description": "The N of \"Speaker N\" in get_transcription's text, or speakerNumber in its segments. Works while a live transcription is still running.",
478
+ "description": "With transcriptionId: the N of \"Speaker N\" in get_transcription's text, or speakerNumber in its segments; it stays the same after naming.",
517
479
  "examples": [
518
480
  2
519
481
  ],
520
482
  "minimum": 1,
521
483
  "type": "integer"
522
484
  },
485
+ "speakerId": {
486
+ "description": "Instead of transcriptionId and speaker: a saved speaker's id (id in list_speakers, or speakerId from save_speaker), changed everywhere they are recognised.",
487
+ "examples": [
488
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
489
+ ],
490
+ "format": "uuid",
491
+ "type": "string"
492
+ },
523
493
  "transcriptionId": {
524
- "description": "A transcriptionId from transcribe, list_transcriptions, search_transcriptions or start_broadcast.",
494
+ "description": "With speaker: the transcription the speaker is in (from create_transcription, list_transcriptions, search_transcriptions or start_broadcast). Leave out with speakerId.",
525
495
  "examples": [
526
496
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
527
497
  ],
@@ -529,25 +499,20 @@
529
499
  "type": "string"
530
500
  }
531
501
  },
532
- "required": [
533
- "transcriptionId",
534
- "speaker",
535
- "name"
536
- ],
537
502
  "type": "object"
538
503
  },
539
504
  "annotations": {
540
- "title": "Save a speaker",
505
+ "title": "Update a speaker",
541
506
  "readOnlyHint": false,
542
507
  "destructiveHint": false,
543
- "idempotentHint": false,
508
+ "idempotentHint": true,
544
509
  "openWorldHint": false
545
510
  }
546
511
  },
547
512
  {
548
513
  "name": "delete_speaker",
549
514
  "title": "Delete a saved speaker",
550
- "description": "Delete a saved speaker for good: their name, fingerprint, stored sample and any cloned voice are erased at once and cannot be restored.\nUse: the person should no longer be recognised, or asked for their voice to be removed. A name inside one transcription: rename_speaker.\nBehaviour: free. Transcripts keep every line they spoke; lines labelled with their saved name go back to \"Speaker N\".\nReturns: speakerId, deleted true.\nErrors: not_found for an unknown or already deleted id; forbidden for a read-only key.",
515
+ "description": "Delete a saved speaker for good: their name, fingerprint, stored sample and any cloned voice are erased at once and cannot be restored.\nUse: the person should no longer be recognised, or asked for their voice to be removed. A name inside one transcription: update_speaker.\nArguments: speakerId is a saved speaker's id, the id field of list_speakers or the speakerId save_speaker returned; not a speaker number (\"Speaker 2\") and not a transcriptionId. To unname someone in one transcription instead: update_speaker with an empty name.\nBehaviour: free. Transcripts keep every line they spoke; lines labelled with their saved name go back to \"Speaker N\".\nReturns: speakerId, deleted true.\nErrors: not_found for an unknown or already deleted id; forbidden for a read-only key.",
551
516
  "inputSchema": {
552
517
  "additionalProperties": false,
553
518
  "properties": {
@@ -576,7 +541,7 @@
576
541
  {
577
542
  "name": "list_broadcasts",
578
543
  "title": "List broadcasts",
579
- "description": "List this account's broadcasts, or get one in detail. A broadcast is a public link where listeners follow a live transcription in their own language.\nUse: to find a broadcastId for start_broadcast or stop_broadcast, see which are on air, or (with broadcastId) a broadcast's recorded transcriptions and streams. Text: get_transcription; they also appear in list_transcriptions.\nBehaviour: free; most recently updated first, up to 500 per lifecycle, no paging.\nReturns: broadcasts (broadcastId, name, link, access, onAir, listenersNow, lastRecordedAt once it has recorded, archived for one in the trash; while on air transcriptionId, state, onAirSince). With broadcastId, one item that adds transcriptions (up to 20: transcriptionId, name, recording, lastRecordedAt) and streams (name, url). None: an empty list.\nErrors: not_found for an unknown broadcastId.",
544
+ "description": "List this account's broadcasts, or get one in detail. A broadcast is a public link where listeners follow a live transcription in their own language.\nUse: to find a broadcastId for start_broadcast, stop_broadcast or delete_broadcast, see which are on air, or (with broadcastId) a broadcast's recorded transcriptions and streams. Text: get_transcription; they also appear in list_transcriptions.\nBehaviour: free; most recently updated first, up to 500 per lifecycle, no paging.\nReturns: broadcasts (broadcastId, name, link, access, onAir, listenersNow, lastRecordedAt once it has recorded, archived for one in the trash; while on air transcriptionId, state, onAirSince). With broadcastId, one item that adds transcriptions (up to 20: transcriptionId, name, recording, lastRecordedAt) and streams (name, url). None: an empty list.\nErrors: not_found for an unknown broadcastId.",
580
545
  "inputSchema": {
581
546
  "additionalProperties": false,
582
547
  "properties": {
@@ -611,7 +576,7 @@
611
576
  {
612
577
  "name": "start_broadcast",
613
578
  "title": "Start a broadcast",
614
- "description": "Start live transcription and translation of a stream playing now, on a public link listeners open in their own language.\nUse: a live stream (YouTube live, internet radio, HLS, Icecast). A recording that has ended: transcribe. Reading it while it runs: get_transcription.\nList price: $0.60 an hour while it runs, billed per second; Pay as you go bills at least 5 minutes per transcription. Listeners and translations are free. It runs until stop_broadcast or the stream ends.\nTranscriptions running at once per account: 5 on Pay as you go, 50 on Premium; past that a start is refused with quota_exceeded.\nBehaviour: each call starts a new run unless it repeats an idempotencyKey; omitting broadcastId makes a new broadcast with a new link.\nReturns: broadcastId, name, link (anyone who opens it reads, or with spokenTranslations also hears, the stream in their language), transcriptionId, state.\nErrors: invalid_state (with link) when already on air, invalid_options for a non-http(s) address, insufficient_credit (with topUpUrl), quota_exceeded, not_found for an unknown broadcastId, forbidden for a key without the write and recordings:publish scopes.",
579
+ "description": "Start live transcription and translation of a stream playing now, on a public link listeners open in their own language.\nUse: a stream playing now. A recording that has ended: create_transcription. Reading it while it runs: get_transcription.\nArguments: streamUrl must be live when called; the same streamUrl on the same broadcast reuses its stream. broadcastId (from list_broadcasts) reruns an existing broadcast on its old link, and then name is ignored; without it a new broadcast is made, named name or the stream's host. spokenLanguage is a code like en (omit to detect). spokenTranslations true adds audio of each translation. idempotencyKey: any string you choose, reused only to retry this same call.\nList price: $0.60 an hour while it runs, billed per second; Pay as you go bills at least 5 minutes per transcription. Listeners and translations are free. It runs until stop_broadcast or the stream ends.\nTranscriptions running at once per account: 5 on Pay as you go, 50 on Premium; past that a start is refused with quota_exceeded.\nBehaviour: each call starts a new run unless it repeats an idempotencyKey; omitting broadcastId makes a new broadcast with a new link.\nReturns: broadcastId, name, link (anyone who opens it reads, or with spokenTranslations also hears, the stream in their language), transcriptionId, state.\nErrors: invalid_state (with link) when already on air, invalid_options for a non-http(s) address, insufficient_credit (with topUpUrl), quota_exceeded, not_found for an unknown broadcastId, forbidden for a key without the write and recordings:publish scopes.",
615
580
  "inputSchema": {
616
581
  "additionalProperties": false,
617
582
  "properties": {
@@ -652,7 +617,7 @@
652
617
  "type": "boolean"
653
618
  },
654
619
  "streamUrl": {
655
- "description": "http(s) address of a stream playing now: a YouTube live, an internet radio station, an HLS (.m3u8) or Icecast stream. A recording that has ended goes to transcribe instead.",
620
+ "description": "http(s) address of a stream playing now: a YouTube live, an internet radio station, an HLS (.m3u8) or Icecast stream. A recording that has ended goes to create_transcription instead.",
656
621
  "examples": [
657
622
  "https://www.youtube.com/watch?v=jfKfPfyJRdk",
658
623
  "https://stream.example.org/radio.m3u8"
@@ -675,10 +640,85 @@
675
640
  "openWorldHint": true
676
641
  }
677
642
  },
643
+ {
644
+ "name": "update_broadcast",
645
+ "title": "Update a broadcast",
646
+ "description": "Change a broadcast's settings: its name, who can listen (public, or private with a password), the language spoken, and whether translations are read aloud, in stock or the speakers' own cloned voices.\nUse: before or between runs; settings apply from the next start_broadcast, the name and access at once. Only the fields given change; give at least one. Starting it: start_broadcast; ending it: stop_broadcast; removing it: delete_broadcast.\nArguments: broadcastId (required) from list_broadcasts or start_broadcast. name 1 to 120 characters. access private needs a password (at least 4 characters) the first time; a new password replaces the old one. spokenLanguage a code like en, or auto. spokenTranslations true or false. clonedVoices true needs consent true, recording that everyone speaking agreed to be cloned.\nBehaviour: free; the link stays the same. Changing access or the password signs out every current listener, who must open the link again. Calling again with the same fields changes nothing else.\nReturns: the broadcast (broadcastId, name, link, access, onAir, listenersNow) with spokenLanguage, spokenTranslations and clonedVoices.\nErrors: invalid_options for nothing to change, a bad name, a short password, private without a password, an unknown language, or clonedVoices true without consent; invalid_state for a broadcast in the trash; not_found for an unknown broadcastId; forbidden for a read-only key.",
647
+ "inputSchema": {
648
+ "additionalProperties": false,
649
+ "properties": {
650
+ "access": {
651
+ "description": "public: anyone with the link can follow it; private: listeners must enter password. Changing it signs out every current listener.",
652
+ "enum": [
653
+ "public",
654
+ "private"
655
+ ]
656
+ },
657
+ "broadcastId": {
658
+ "description": "A broadcastId from list_broadcasts or start_broadcast.",
659
+ "examples": [
660
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
661
+ ],
662
+ "format": "uuid",
663
+ "type": "string"
664
+ },
665
+ "clonedVoices": {
666
+ "description": "With spokenTranslations: true reads each speaker's translated words in that speaker's own cloned voice (needs consent true); false: stock voices.",
667
+ "type": "boolean"
668
+ },
669
+ "consent": {
670
+ "default": false,
671
+ "description": "Must be true with clonedVoices true: confirms everyone speaking on it agreed to have their voice cloned. Recorded with your account and the time. Not needed to turn cloned voices off.",
672
+ "type": "boolean"
673
+ },
674
+ "name": {
675
+ "description": "New name shown to listeners; trimmed, 1 to 120 characters.",
676
+ "examples": [
677
+ "Sunday service"
678
+ ],
679
+ "maxLength": 120,
680
+ "minLength": 1,
681
+ "type": "string"
682
+ },
683
+ "password": {
684
+ "description": "The listeners' password, at least 4 characters: required to make a broadcast private the first time, and setting a new one signs out every current listener. Ignored with access public.",
685
+ "examples": [
686
+ "choir2026"
687
+ ],
688
+ "maxLength": 200,
689
+ "minLength": 4,
690
+ "type": "string"
691
+ },
692
+ "spokenLanguage": {
693
+ "description": "Code of the language spoken (e.g. en, es), or auto to detect it. Applies from the next start_broadcast.",
694
+ "examples": [
695
+ "en",
696
+ "auto"
697
+ ],
698
+ "type": "string"
699
+ },
700
+ "spokenTranslations": {
701
+ "description": "true: each translation is also read aloud, so listeners can hear it in their language; false: text only. Applies from the next start_broadcast. Listening stays free.",
702
+ "type": "boolean"
703
+ }
704
+ },
705
+ "required": [
706
+ "broadcastId"
707
+ ],
708
+ "type": "object"
709
+ },
710
+ "annotations": {
711
+ "title": "Update a broadcast",
712
+ "readOnlyHint": false,
713
+ "destructiveHint": false,
714
+ "idempotentHint": true,
715
+ "openWorldHint": false
716
+ }
717
+ },
678
718
  {
679
719
  "name": "stop_broadcast",
680
720
  "title": "Stop a broadcast",
681
- "description": "Stop the live transcription a broadcast is running, which also stops its billing.\nUse: when the event is over. To remove the transcript afterwards: delete_transcription.\nBehaviour: free. The transcript is kept (get_transcription and get_summary work on it once state is done); the broadcastId and its link stay, ready for the next start_broadcast.\nReturns: the stopped transcription's status (transcriptionId, name, kind live, state finishing then done, startedAt, endedAt, durationS) with broadcastId and link.\nErrors: invalid_state when not on air, not_found for an unknown broadcastId, forbidden for a key without the recordings:publish scope.",
721
+ "description": "Stop the live transcription a broadcast is running, which also stops its billing.\nUse: when the event is over. To remove the transcript afterwards: delete_transcription; the broadcast and its link: delete_broadcast.\nArguments: broadcastId of a broadcast on air: from start_broadcast's result, or list_broadcasts (onAir true); not the transcriptionId. A second call is refused with invalid_state and changes nothing, so a retry is safe.\nBehaviour: free. The transcript is kept (get_transcription reads it, and its summary, once state is done); the broadcastId and its link stay, ready for the next start_broadcast.\nReturns: the stopped transcription's status (transcriptionId, name, kind live, state finishing then done, startedAt, endedAt, durationS) with broadcastId and link.\nErrors: invalid_state when not on air, not_found for an unknown broadcastId, forbidden for a key without the recordings:publish scope.",
682
722
  "inputSchema": {
683
723
  "additionalProperties": false,
684
724
  "properties": {
@@ -703,6 +743,35 @@
703
743
  "idempotentHint": true,
704
744
  "openWorldHint": false
705
745
  }
746
+ },
747
+ {
748
+ "name": "delete_broadcast",
749
+ "title": "Delete a broadcast",
750
+ "description": "Move a broadcast to the trash: its public link and stream addresses stop working for listeners.\nUse: a broadcast that will not run again. For 30 days it can be restored from the trash in the VocalStack Dashboard (list_broadcasts with lifecycle archived lists it); after that the broadcast is erased for good. Ending a live run: stop_broadcast.\nArguments: broadcastId from list_broadcasts or start_broadcast; not a transcriptionId.\nBehaviour: free. A broadcast on air is refused: stop it first (stop_broadcast). The transcriptions it recorded are not deleted: while it is in the trash they leave list_transcriptions with it, get_transcription still reads them by id, and delete_transcription deletes one. Deleting again changes nothing.\nReturns: broadcastId, name, deleted true, deletedAt, restorableUntil.\nErrors: invalid_state while on air; not_found for an unknown broadcastId; forbidden for a read-only key.",
751
+ "inputSchema": {
752
+ "additionalProperties": false,
753
+ "properties": {
754
+ "broadcastId": {
755
+ "description": "A broadcastId from list_broadcasts or start_broadcast, of a broadcast that is not on air.",
756
+ "examples": [
757
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
758
+ ],
759
+ "format": "uuid",
760
+ "type": "string"
761
+ }
762
+ },
763
+ "required": [
764
+ "broadcastId"
765
+ ],
766
+ "type": "object"
767
+ },
768
+ "annotations": {
769
+ "title": "Delete a broadcast",
770
+ "readOnlyHint": false,
771
+ "destructiveHint": true,
772
+ "idempotentHint": true,
773
+ "openWorldHint": false
774
+ }
706
775
  }
707
776
  ]
708
777
  }
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@vocalstack/mcp",
3
- "version": "1.3.0",
3
+ "version": "1.5.0",
4
4
  "description": "VocalStack's MCP server: transcribe, translate, search and summarize audio and video, and run live broadcasts, from any MCP client, on VocalStack's public API.",
5
5
  "license": "MIT",
6
6
  "type": "module",