@vocalstack/mcp 1.2.0 → 1.4.0

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/dist/tools.json CHANGED
@@ -1,104 +1,15 @@
1
1
  {
2
- "instructions": "VocalStack transcribes and translates recorded or live audio and video in about 100 languages, with speaker labels. Recorded: transcribe_url (a public link) or transcribe_file (base64, up to 10 MiB) returns a transcriptionId at once; wait_for_transcription until state is \"done\" (call again while timedOut is true); get_transcription reads the text (next page: offset = nextOffset). Find: list_transcriptions (ids, newest first), search_transcriptions (exact words or names), search_by_meaning (a question, any language). One transcription: get_summary (overview), translate_transcription (whole text in another language), rename_speaker (name \"Speaker N\"), set_cloned_voices (spoken translations in each speaker's own voice, with their consent), delete_transcription (to the trash, restorable for 30 days). Live: start_broadcast transcribes a live stream on a public link where listeners follow it in their own language; get_transcription reads it as it runs; stop_broadcast ends it; list_broadcasts and get_broadcast show broadcasts. Reference: list_voices (voices per language), list_speakers (saved people). Cost: only transcription is billed, per second of audio; everything else, and listening, is free. Retry a start with the same idempotencyKey. Errors are tool results with isError true: error (not_found, invalid_options, invalid_state, forbidden, insufficient_credit, rate_limited), 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_summary; rename_speaker; set_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; delete_broadcast removes one.\nPeople: list_speakers, save_speaker (voice cloned only with the person's consent), rename_speaker, set_cloned_voices with speakerId, 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_url",
6
- "title": "Transcribe a link",
7
- "description": "Start transcribing a recorded audio or video at a public link: a media file, a podcast episode, a YouTube video, or a public post on one of the sites listed under url. Use transcribe_file for a file you hold (up to 10 MiB) and start_broadcast for a live stream, which this tool refuses. Billed per second of audio transcribed (only startS to endS when set); a refused call costs nothing. Each call creates and bills a new transcription unless it repeats an idempotencyKey. Returns at once, before any text: transcriptionId, state (\"queued\", then \"fetching\", \"transcribing\", and finally \"done\", \"failed\" or \"stopped\"), progress, and shareUrl when shareLink is true; next call wait_for_transcription. Errors: invalid_options (not a public http(s) link), insufficient_credit (with topUpUrl), quota_exceeded, rate_limited; a link that cannot be read ends in state \"failed\" with error.",
5
+ "name": "create_transcription",
6
+ "title": "Create a transcription",
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,
10
10
  "properties": {
11
11
  "endS": {
12
- "description": "Transcribe only up to this second (default: the end); at least 0.5 s after startS. Only the part transcribed is billed.",
13
- "examples": [
14
- 600
15
- ],
16
- "type": "number"
17
- },
18
- "idempotencyKey": {
19
- "description": "Any unique string of visible ASCII. Send the same key with the same arguments when retrying: for 24 h the first result comes back (with replayed true) and nothing is created or billed twice.",
20
- "examples": [
21
- "meeting-2026-11-19"
22
- ],
23
- "maxLength": 255,
24
- "type": "string"
25
- },
26
- "name": {
27
- "description": "Display name shown in lists and on the share page. Defaults to the file name or the link.",
28
- "examples": [
29
- "Weekly team call"
30
- ],
31
- "maxLength": 200,
32
- "type": "string"
33
- },
34
- "shareLink": {
35
- "default": false,
36
- "description": "Make a public link anyone can open to read the transcript, without signing in (shareUrl in the result). Off (the default): the transcription stays private to this account.",
37
- "type": "boolean"
38
- },
39
- "speakerLabels": {
40
- "default": "auto",
41
- "description": "Label who is speaking: auto (the default) labels lines once a second speaker is heard, on labels every line, off none. Recordings under a minute are never labelled.",
42
- "enum": [
43
- "auto",
44
- "on",
45
- "off"
46
- ]
47
- },
48
- "spokenLanguage": {
49
- "description": "Code of the language spoken, e.g. en, ro, es. Omit to detect it (recordings that switch language are handled too); set it when you know it, for better accuracy on short or noisy audio.",
50
- "examples": [
51
- "en",
52
- "es"
53
- ],
54
- "type": "string"
55
- },
56
- "startS": {
57
- "description": "Transcribe only from this second of the recording (default: the start). Only the part transcribed is billed.",
58
- "examples": [
59
- 90
60
- ],
61
- "minimum": 0,
62
- "type": "number"
63
- },
64
- "summary": {
65
- "default": false,
66
- "description": "Also write a title and summary once it is done, at no extra cost (get_summary reads it; asking get_summary later writes one anyway). Never for a transcription under a minute.",
67
- "type": "boolean"
68
- },
69
- "url": {
70
- "description": "Public http(s) link to an audio or video file, a YouTube video, or a public recording on Twitch, Facebook, Instagram, Vimeo, Dailymotion, TikTok, X, SoundCloud, Kick, Mixcloud, Bandcamp, LinkedIn, TED, Internet Archive, Streamable, Odysee, Niconico, VK, BitChute, Wistia, Vidyard, Kaltura, JW Player, Pinterest, Bluesky, Google Drive, Dropbox, Spreaker, iHeartRadio, Audioboom, CNN or Loom. It must open without signing in. Live streams are not accepted here (use start_broadcast).",
71
- "examples": [
72
- "https://www.youtube.com/watch?v=jNQXAC9IVRw",
73
- "https://example.com/episode-12.mp3"
74
- ],
75
- "format": "uri",
76
- "maxLength": 2048,
77
- "type": "string"
78
- }
79
- },
80
- "required": [
81
- "url"
82
- ],
83
- "type": "object"
84
- },
85
- "annotations": {
86
- "title": "Transcribe a link",
87
- "readOnlyHint": false,
88
- "destructiveHint": false,
89
- "idempotentHint": false,
90
- "openWorldHint": true
91
- }
92
- },
93
- {
94
- "name": "transcribe_file",
95
- "title": "Transcribe a file",
96
- "description": "Start transcribing an audio or video file you hold, sent whole as base64 (at most 10 MiB decoded; mp3, m4a, wav, mp4, webm and other common formats). Use transcribe_url for a larger file or one already online, and start_broadcast for a live stream. Billed per second of audio transcribed (only startS to endS when set); a refused call costs nothing. Each call creates and bills a new transcription unless it repeats an idempotencyKey. Returns at once, before any text: transcriptionId, state (\"queued\", \"transcribing\", then \"done\", \"failed\" or \"stopped\"), progress, and shareUrl when shareLink is true; next call wait_for_transcription. The transcriptionId stays valid until the transcription is deleted. Errors: invalid_options (bad base64, over 10 MiB), invalid_audio (empty file), insufficient_credit (with topUpUrl), quota_exceeded, rate_limited; audio that cannot be decoded ends in state \"failed\" with error.",
97
- "inputSchema": {
98
- "additionalProperties": false,
99
- "properties": {
100
- "endS": {
101
- "description": "Transcribe only up to this second (default: the end); at least 0.5 s after startS. Only the part transcribed is billed.",
12
+ "description": "Transcribe only up to this second (default: the end), at least 0.5 s after startS.",
102
13
  "examples": [
103
14
  600
104
15
  ],
@@ -106,11 +17,11 @@
106
17
  },
107
18
  "fileBase64": {
108
19
  "contentEncoding": "base64",
109
- "description": "The whole file, standard base64 (a data: URL prefix and line breaks are accepted). At most 10 MiB decoded; for a larger file use transcribe_url with a link to it.",
20
+ "description": "The whole file in standard base64 (a data: URL prefix and line breaks are accepted), at most 10 MiB decoded; for a larger file give url instead. mp3, m4a, wav, mp4, webm and other common formats; the format is read from the file. Needs filename.",
110
21
  "type": "string"
111
22
  },
112
23
  "filename": {
113
- "description": "The file's name, with its extension; also the default display name.",
24
+ "description": "With fileBase64 only: the file's name with its extension, also the default display name.",
114
25
  "examples": [
115
26
  "meeting.m4a"
116
27
  ],
@@ -118,7 +29,7 @@
118
29
  "type": "string"
119
30
  },
120
31
  "idempotencyKey": {
121
- "description": "Any unique string of visible ASCII. Send the same key with the same arguments when retrying: for 24 h the first result comes back (with replayed true) and nothing is created or billed twice.",
32
+ "description": "Any unique string of visible ASCII. Retrying with the same key and the same arguments within 24 h returns the first result (with replayed true) and nothing is created or billed twice; the same key with other arguments is refused.",
122
33
  "examples": [
123
34
  "meeting-2026-11-19"
124
35
  ],
@@ -126,7 +37,7 @@
126
37
  "type": "string"
127
38
  },
128
39
  "mimeType": {
129
- "description": "The file's media type, when known. Optional: the format is read from the file itself.",
40
+ "description": "With fileBase64 only, optional: the file's media type when known.",
130
41
  "examples": [
131
42
  "audio/mpeg",
132
43
  "video/mp4"
@@ -143,12 +54,12 @@
143
54
  },
144
55
  "shareLink": {
145
56
  "default": false,
146
- "description": "Make a public link anyone can open to read the transcript, without signing in (shareUrl in the result). Off (the default): the transcription stays private to this account.",
57
+ "description": "Make a public link anyone can open to read the transcript without signing in (shareUrl in the result). Off: it stays private to this account.",
147
58
  "type": "boolean"
148
59
  },
149
60
  "speakerLabels": {
150
61
  "default": "auto",
151
- "description": "Label who is speaking: auto (the default) labels lines once a second speaker is heard, on labels every line, off none. Recordings under a minute are never labelled.",
62
+ "description": "auto labels lines once a second speaker is heard, on labels every line, off none. Recordings under a minute are never labelled.",
152
63
  "enum": [
153
64
  "auto",
154
65
  "on",
@@ -156,7 +67,7 @@
156
67
  ]
157
68
  },
158
69
  "spokenLanguage": {
159
- "description": "Code of the language spoken, e.g. en, ro, es. Omit to detect it (recordings that switch language are handled too); set it when you know it, for better accuracy on short or noisy audio.",
70
+ "description": "Code of the language spoken, e.g. en, ro, es. Omit to detect it (recordings that switch language are handled too); set it for better accuracy on short or noisy audio.",
160
71
  "examples": [
161
72
  "en",
162
73
  "es"
@@ -164,7 +75,7 @@
164
75
  "type": "string"
165
76
  },
166
77
  "startS": {
167
- "description": "Transcribe only from this second of the recording (default: the start). Only the part transcribed is billed.",
78
+ "description": "Transcribe only from this second (default: the start). Only the part transcribed is billed: startS 90 and endS 600 bill 8.5 minutes.",
168
79
  "examples": [
169
80
  90
170
81
  ],
@@ -173,34 +84,40 @@
173
84
  },
174
85
  "summary": {
175
86
  "default": false,
176
- "description": "Also write a title and summary once it is done, at no extra cost (get_summary reads it; asking get_summary later writes one anyway). Never for a transcription under a minute.",
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.",
177
88
  "type": "boolean"
89
+ },
90
+ "url": {
91
+ "description": "Public http(s) link to the recording: an audio or video file, a YouTube video, or a public post on Twitch, Facebook, Instagram, Vimeo, Dailymotion, TikTok, X, SoundCloud, Kick, Mixcloud, Bandcamp, LinkedIn, TED, Internet Archive, Streamable, Odysee, Niconico, VK, BitChute, Wistia, Vidyard, Kaltura, JW Player, Pinterest, Bluesky, Google Drive, Dropbox, Spreaker, iHeartRadio, Audioboom, CNN or Loom. It must open without signing in and must not be a live stream. Give either url or fileBase64, not both.",
92
+ "examples": [
93
+ "https://www.youtube.com/watch?v=jNQXAC9IVRw",
94
+ "https://example.com/episode-12.mp3"
95
+ ],
96
+ "format": "uri",
97
+ "maxLength": 2048,
98
+ "type": "string"
178
99
  }
179
100
  },
180
- "required": [
181
- "filename",
182
- "fileBase64"
183
- ],
184
101
  "type": "object"
185
102
  },
186
103
  "annotations": {
187
- "title": "Transcribe a file",
104
+ "title": "Create a transcription",
188
105
  "readOnlyHint": false,
189
106
  "destructiveHint": false,
190
107
  "idempotentHint": false,
191
- "openWorldHint": false
108
+ "openWorldHint": true
192
109
  }
193
110
  },
194
111
  {
195
112
  "name": "get_transcription",
196
113
  "title": "Read a transcription",
197
- "description": "Read one transcription's text, with timestamps and speakers, and its status. Use it once wait_for_transcription reports \"done\", or at any time on a live transcription for the text so far. For another language use translate_transcription, for an overview get_summary, to find a passage across transcriptions search_transcriptions. Read-only and free. format \"text\" (default) gives lines like \"[mm:ss] Speaker 1: words\"; \"segments\" gives start, end, text, speakerNumber and speakerName per line (each word's timing too with includeWords); \"none\" gives the status only. Returns status (state, progress, error when it failed), segmentCount, and the text or segments; partial is true while text is still coming. Long transcripts come in pages of limit segments (default 500): when the result has nextOffset, call again with offset set to that value. Unknown id: not_found.",
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 create_transcription (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, shareUrl when it has a public link: made with create_transcription's shareLink, or in the Dashboard), 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).",
198
115
  "inputSchema": {
199
116
  "additionalProperties": false,
200
117
  "properties": {
201
118
  "format": {
202
119
  "default": "text",
203
- "description": "text (the default): one readable string, a line per segment as \"[mm:ss] Speaker 1: words\", best for reading or quoting; segments: an array of timed lines, best for subtitles or exact times; none: status only, no 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).",
204
121
  "enum": [
205
122
  "text",
206
123
  "segments",
@@ -209,12 +126,20 @@
209
126
  },
210
127
  "includeWords": {
211
128
  "default": false,
212
- "description": "With format segments, also return each word with its start and end second. Off by default: it makes the answer several times larger.",
129
+ "description": "With format segments and no lang: also each word with its start and end second. It makes the answer several times larger, so ask for a narrow page (e.g. limit 50).",
213
130
  "type": "boolean"
214
131
  },
132
+ "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
+ "examples": [
135
+ "es",
136
+ "ja"
137
+ ],
138
+ "type": "string"
139
+ },
215
140
  "limit": {
216
141
  "default": 500,
217
- "description": "Most segments to return in one call (default 500). When more are left the result has nextOffset; ask again with offset set to it.",
142
+ "description": "Most segments in one call. When more are left the result has nextOffset.",
218
143
  "examples": [
219
144
  200
220
145
  ],
@@ -224,7 +149,7 @@
224
149
  },
225
150
  "offset": {
226
151
  "default": 0,
227
- "description": "Index of the first segment to return (default 0, the start). To page, pass the nextOffset the previous call returned.",
152
+ "description": "Index of the first segment to return (0, the start, by default). To page, pass the nextOffset the previous call returned.",
228
153
  "examples": [
229
154
  500
230
155
  ],
@@ -232,51 +157,22 @@
232
157
  "type": "integer"
233
158
  },
234
159
  "transcriptionId": {
235
- "description": "The transcriptionId that transcribe_url, transcribe_file, list_transcriptions, a search tool or start_broadcast returned.",
160
+ "description": "A transcriptionId from create_transcription, list_transcriptions, search_transcriptions or start_broadcast.",
236
161
  "examples": [
237
162
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
238
163
  ],
239
164
  "format": "uuid",
240
165
  "type": "string"
241
- }
242
- },
243
- "required": [
244
- "transcriptionId"
245
- ],
246
- "type": "object"
247
- },
248
- "annotations": {
249
- "title": "Read a transcription",
250
- "readOnlyHint": true,
251
- "destructiveHint": false,
252
- "idempotentHint": true,
253
- "openWorldHint": false
254
- }
255
- },
256
- {
257
- "name": "wait_for_transcription",
258
- "title": "Wait for a transcription",
259
- "description": "Wait until a file or link transcription finishes, up to timeoutS seconds (default 50, at most 60), and return its status without the text. Use it after transcribe_url or transcribe_file instead of calling get_transcription in a loop. Read-only and free; clients that send a progress token get progress notifications. Returns transcriptionId, state, progress (pct, processedS, totalS), error when it failed, and timedOut. timedOut true means it is still running: call again. State \"done\": read the text with get_transcription; \"failed\": error says why. A live transcription runs until stopped, so for one this returns its status at once. Errors: not_found; rate_limited past 4 open waits per key (8 per account).",
260
- "inputSchema": {
261
- "additionalProperties": false,
262
- "properties": {
263
- "timeoutS": {
264
- "default": 50,
265
- "description": "Longest to wait in this call, in seconds (default 50, at most 60). It returns sooner when the transcription finishes; otherwise it returns with timedOut true and you call it again. Lower it if your client gives up on calls sooner.",
166
+ },
167
+ "waitS": {
168
+ "default": 0,
169
+ "description": "Seconds to wait for a file or link transcription to finish before answering, 0 to 60 (0, the default: answer at once). Returns as soon as it ends; timedOut true means it is still running, so call again. Keep it under your client's call timeout. A live transcription never waits.",
266
170
  "examples": [
267
- 30
171
+ 50
268
172
  ],
269
173
  "maximum": 60,
270
- "minimum": 1,
174
+ "minimum": 0,
271
175
  "type": "integer"
272
- },
273
- "transcriptionId": {
274
- "description": "The transcriptionId that transcribe_url, transcribe_file, list_transcriptions, a search tool or start_broadcast returned.",
275
- "examples": [
276
- "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
277
- ],
278
- "format": "uuid",
279
- "type": "string"
280
176
  }
281
177
  },
282
178
  "required": [
@@ -285,7 +181,7 @@
285
181
  "type": "object"
286
182
  },
287
183
  "annotations": {
288
- "title": "Wait for a transcription",
184
+ "title": "Read a transcription",
289
185
  "readOnlyHint": true,
290
186
  "destructiveHint": false,
291
187
  "idempotentHint": true,
@@ -295,17 +191,17 @@
295
191
  {
296
192
  "name": "list_transcriptions",
297
193
  "title": "List transcriptions",
298
- "description": "List this account's transcriptions (files, links and live ones, including those recorded by broadcasts), newest first, one page at a time. Use it to find a transcriptionId or see what exists; to find what was said use search_transcriptions (exact words) or search_by_meaning (a question); for broadcasts use list_broadcasts. Read-only and free. lifecycle: active (default), archived (deleted, restorable for 30 days) or all. Returns transcriptions, each with transcriptionId, name, createdAt, kind (\"file\" or \"live\"), state, durationS and title when known, and broadcastId for a broadcast's live transcription; plus nextCursor when more remain: pass it back as cursor. None: an empty list.",
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: 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.",
299
195
  "inputSchema": {
300
196
  "additionalProperties": false,
301
197
  "properties": {
302
198
  "cursor": {
303
- "description": "Omit for the first page (the newest). For the next page, pass the nextCursor the previous call returned, unchanged.",
199
+ "description": "Omit for the first page (the newest). For the next page, pass the previous call's nextCursor unchanged, with the same lifecycle.",
304
200
  "type": "string"
305
201
  },
306
202
  "lifecycle": {
307
203
  "default": "active",
308
- "description": "Which transcriptions to list: active (the default) those in use; archived those in the trash; all both.",
204
+ "description": "Which transcriptions: active those in use, archived those in the trash, all both.",
309
205
  "enum": [
310
206
  "active",
311
207
  "archived",
@@ -314,7 +210,7 @@
314
210
  },
315
211
  "limit": {
316
212
  "default": 20,
317
- "description": "Most transcriptions in one page (default 20, at most 100). nextCursor in the result means there are more.",
213
+ "description": "Most transcriptions in one page.",
318
214
  "examples": [
319
215
  50
320
216
  ],
@@ -335,13 +231,13 @@
335
231
  },
336
232
  {
337
233
  "name": "search_transcriptions",
338
- "title": "Search for words",
339
- "description": "Find the lines where given words or a name were said, across all of this account's transcriptions or in one (transcriptionId). It matches the words themselves, allowing close spellings, and also searches the query's English translation. For a question or topic whose exact words you do not know use search_by_meaning; to read around a hit, get_transcription. Read-only and free. Returns up to limit hits (default 20, at most 50, no next page: narrow the query or set transcriptionId), best first: whole-word matches, then closest spelling, then newest. Each hit has transcriptionId, name, mediaStart (the second it was said), speaker, text, score, live and recordedAt, plus rendered (the line translated into lang) when lang is set; lang never changes what is searched. Also returns query and queryEn. No match: an empty hits list. Unknown transcriptionId: not_found.",
234
+ "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.",
340
236
  "inputSchema": {
341
237
  "additionalProperties": false,
342
238
  "properties": {
343
239
  "lang": {
344
- "description": "Optional language code, e.g. en: each hit also comes back translated into it, as rendered. It never changes what is searched or matched.",
240
+ "description": "mode words only, optional language code: each hit also comes back translated into it, as rendered. It never changes what matches.",
345
241
  "examples": [
346
242
  "en",
347
243
  "de"
@@ -350,7 +246,7 @@
350
246
  },
351
247
  "limit": {
352
248
  "default": 20,
353
- "description": "Most matching lines to return, best first (default 20, at most 50). There is no next page: narrow the query or set transcriptionId instead.",
249
+ "description": "mode words only: most lines to return, best first. There is no next page: narrow the query or set transcriptionId.",
354
250
  "examples": [
355
251
  10
356
252
  ],
@@ -358,57 +254,26 @@
358
254
  "minimum": 1,
359
255
  "type": "integer"
360
256
  },
361
- "query": {
362
- "description": "Words, a phrase or a name that was said (2 to 200 characters), in any language.",
363
- "examples": [
364
- "quarterly budget",
365
- "María"
366
- ],
367
- "maxLength": 200,
368
- "minLength": 2,
369
- "type": "string"
257
+ "mode": {
258
+ "default": "words",
259
+ "description": "words: the lines where these words or a close spelling were said (\"presupuesto\" also finds \"budget\"). meaning: the passages closest in meaning, even with none of the words.",
260
+ "enum": [
261
+ "words",
262
+ "meaning"
263
+ ]
370
264
  },
371
- "transcriptionId": {
372
- "description": "Optional: search only this transcription (a transcriptionId from list_transcriptions or a transcribe tool). Omit to search all of this account's transcriptions.",
373
- "examples": [
374
- "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
375
- ],
376
- "format": "uuid",
377
- "type": "string"
378
- }
379
- },
380
- "required": [
381
- "query"
382
- ],
383
- "type": "object"
384
- },
385
- "annotations": {
386
- "title": "Search for words",
387
- "readOnlyHint": true,
388
- "destructiveHint": false,
389
- "idempotentHint": true,
390
- "openWorldHint": false
391
- }
392
- },
393
- {
394
- "name": "search_by_meaning",
395
- "title": "Search by meaning",
396
- "description": "Find the passages closest in meaning to a question or topic, across this account's transcriptions or in one (transcriptionId), in any language and even when they share none of its words. Use it to answer questions about what was said; use search_transcriptions for exact words or a name, and get_summary for an overview of one transcription. Read-only and free; limited per hour on some plans (searchesLeftThisHour; past it, rate_limited). Returns up to 10 passages, closest first, each with transcriptionId, name, mediaStart and mediaEnd (seconds), text, similarity (0 to 1) and live; state is \"ready\", or \"unavailable\"/\"off\" with no passages when this search cannot run. Nothing close: an empty list. Cite the transcription and second when answering. Unknown transcriptionId: not_found.",
397
- "inputSchema": {
398
- "additionalProperties": false,
399
- "properties": {
400
265
  "query": {
401
- "description": "A question or topic in plain words (2 to 200 characters), in any language.",
266
+ "description": "mode words: words, a phrase or a name that was said. mode meaning: a question or topic in plain words; a full question works better than one keyword. Any language.",
402
267
  "examples": [
403
- "What did they decide about hiring?",
404
- "pricing complaints"
268
+ "quarterly budget",
269
+ "What did they decide about hiring?"
405
270
  ],
406
271
  "maxLength": 200,
407
272
  "minLength": 2,
408
273
  "type": "string"
409
274
  },
410
275
  "transcriptionId": {
411
- "description": "Optional: search only this transcription (a transcriptionId from list_transcriptions or a transcribe tool). Omit to search all of this account's transcriptions.",
276
+ "description": "Search only this transcription (a transcriptionId from list_transcriptions or create_transcription). Omit to search the whole account.",
412
277
  "examples": [
413
278
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
414
279
  ],
@@ -422,7 +287,7 @@
422
287
  "type": "object"
423
288
  },
424
289
  "annotations": {
425
- "title": "Search by meaning",
290
+ "title": "Search transcriptions",
426
291
  "readOnlyHint": true,
427
292
  "destructiveHint": false,
428
293
  "idempotentHint": true,
@@ -432,19 +297,19 @@
432
297
  {
433
298
  "name": "get_summary",
434
299
  "title": "Get a summary",
435
- "description": "Get the title and summary of a finished transcription. Use it for an overview; get_transcription gives the full text, search_by_meaning answers a specific question, translate_transcription translates the whole text. Free. When there is no summary yet, the first call writes one and keeps it (this needs the write scope; reading an existing one needs only read). Returns state: \"ready\" with title and summary (and lang when translated); \"pending\" while it is being written or the transcription is still running: call again in a few seconds; \"not_requested\" for recordings under a minute, which are never summarised; \"unavailable\" when the transcription failed. lang translates the summary only when the transcription has summary translations on; otherwise the original comes back without lang. Unknown id: not_found.",
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.\nArguments: transcriptionId of a finished transcription (state done, or a stopped live one), from create_transcription, list_transcriptions or search_transcriptions. lang is optional, a language code such as fr: it translates only when the transcription was made with summary translations on and lang differs from the language the summary is in; otherwise the original comes back, without lang in the result.\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.",
436
301
  "inputSchema": {
437
302
  "additionalProperties": false,
438
303
  "properties": {
439
304
  "lang": {
440
- "description": "Optional language code to have the title and summary translated into, when the transcription was made with summary translations on; otherwise the original comes back.",
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.",
441
306
  "examples": [
442
307
  "fr"
443
308
  ],
444
309
  "type": "string"
445
310
  },
446
311
  "transcriptionId": {
447
- "description": "The transcriptionId that transcribe_url, transcribe_file, list_transcriptions, a search tool or start_broadcast returned.",
312
+ "description": "A transcriptionId from create_transcription, list_transcriptions, search_transcriptions or start_broadcast.",
448
313
  "examples": [
449
314
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
450
315
  ],
@@ -466,49 +331,39 @@
466
331
  }
467
332
  },
468
333
  {
469
- "name": "translate_transcription",
470
- "title": "Translate a transcription",
471
- "description": "Get a whole finished transcription translated into another language, in the same text or segments shape as get_transcription. Use get_transcription for the original language, get_summary with lang for a translated overview, and search_transcriptions with lang to translate only matching lines. Needs state \"done\" for a file or link, or a stopped live transcription; earlier it gives invalid_state (read the text so far with get_transcription). Free. The first call for a language can take up to a minute on a long transcription; the translation is kept, so later calls are quick. Returns lang, translated (false when it is already in that language), segmentCount, text or segments (start, end, text, speakerNumber, speakerName), and nextOffset when more remain: call again with offset set to it (limit segments per call, default 500). Unsupported lang: invalid_options; unknown id: not_found.",
334
+ "name": "rename_speaker",
335
+ "title": "Rename a speaker",
336
+ "description": "Name or rename a speaker: one person in one transcription, so its transcript shows the name instead of \"Speaker N\", or a saved speaker, everywhere they are recognised.\nUse: labelling who spoke here, or correcting a saved speaker's name. To recognise the person in later transcriptions too: 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) or speakerId (from list_speakers). name is trimmed; in a transcription it is cut to 80 characters and an empty name undoes it; a saved speaker's name is 1 to 60 characters. Calling again replaces the name.\nBehaviour: free and reversible; in a transcription it applies to that person in every recording of it, and listeners following it live see it at once. The person's speaker number never changes. Renaming a saved speaker changes nothing else about them (fingerprint, consent, cloned voice).\nReturns: transcriptionId, speaker, name (null when cleared); or, with speakerId, speakerId and name.\nErrors: invalid_options for both or neither kind of target, or a saved speaker's name out of 1 to 60; not_found for an unknown transcription, speaker number or speakerId; forbidden for a read-only key.",
472
337
  "inputSchema": {
473
338
  "additionalProperties": false,
474
339
  "properties": {
475
- "format": {
476
- "default": "text",
477
- "description": "text (the default): one readable string, a line per segment as \"[mm:ss] Speaker 1: words\"; segments: an array of timed lines (start, end, speaker, text), best for subtitles.",
478
- "enum": [
479
- "text",
480
- "segments"
481
- ]
482
- },
483
- "lang": {
484
- "description": "Code of the language to translate into, one of about 100 (e.g. es, de, ja, zh, pt).",
340
+ "name": {
341
+ "description": "The name to show instead of \"Speaker N\", trimmed: 1 to 80 characters in one transcription (longer is cut; an empty string clears it back to \"Speaker N\"), 1 to 60 for a saved speaker (never empty).",
485
342
  "examples": [
486
- "es",
487
- "ja"
343
+ "Maria Lopez",
344
+ ""
488
345
  ],
346
+ "maxLength": 80,
489
347
  "type": "string"
490
348
  },
491
- "limit": {
492
- "default": 500,
493
- "description": "Most segments to return in one call (default 500). When more are left the result has nextOffset; ask again with offset set to it.",
349
+ "speaker": {
350
+ "description": "With transcriptionId: the N of \"Speaker N\" in get_transcription's text, or speakerNumber in its segments; it stays the same after naming.",
494
351
  "examples": [
495
- 200
352
+ 2
496
353
  ],
497
- "maximum": 2000,
498
354
  "minimum": 1,
499
355
  "type": "integer"
500
356
  },
501
- "offset": {
502
- "default": 0,
503
- "description": "Index of the first segment to return (default 0, the start). To page, pass the nextOffset the previous call returned.",
357
+ "speakerId": {
358
+ "description": "Instead of transcriptionId and speaker: a saved speaker's id (id in list_speakers, or speakerId from save_speaker), renamed everywhere they are recognised.",
504
359
  "examples": [
505
- 500
360
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
506
361
  ],
507
- "minimum": 0,
508
- "type": "integer"
362
+ "format": "uuid",
363
+ "type": "string"
509
364
  },
510
365
  "transcriptionId": {
511
- "description": "The transcriptionId that transcribe_url, transcribe_file, list_transcriptions, a search tool or start_broadcast returned.",
366
+ "description": "With speaker: the transcription the speaker is in (from create_transcription, list_transcriptions, search_transcriptions or start_broadcast). Leave out with speakerId.",
512
367
  "examples": [
513
368
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
514
369
  ],
@@ -517,13 +372,12 @@
517
372
  }
518
373
  },
519
374
  "required": [
520
- "transcriptionId",
521
- "lang"
375
+ "name"
522
376
  ],
523
377
  "type": "object"
524
378
  },
525
379
  "annotations": {
526
- "title": "Translate a transcription",
380
+ "title": "Rename a speaker",
527
381
  "readOnlyHint": false,
528
382
  "destructiveHint": false,
529
383
  "idempotentHint": true,
@@ -531,30 +385,14 @@
531
385
  }
532
386
  },
533
387
  {
534
- "name": "rename_speaker",
535
- "title": "Rename a speaker",
536
- "description": "Name a speaker in one transcription, so its transcript shows the name instead of \"Speaker N\". speaker is the N that get_transcription shows (speakerNumber in its segments). The name applies to that person in every recording of the transcription, and listeners following it live see it at once. It names them in this transcription only: it does not add them to the account's saved speakers (list_speakers). An empty name clears it back to \"Speaker N\". Free; repeating a call changes nothing. Returns transcriptionId, speaker and name (null when cleared). Errors: not_found for an unknown transcription or a speaker number it does not have; forbidden for a read-only key.",
388
+ "name": "delete_transcription",
389
+ "title": "Delete a transcription",
390
+ "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.",
537
391
  "inputSchema": {
538
392
  "additionalProperties": false,
539
393
  "properties": {
540
- "name": {
541
- "description": "The name to show instead of \"Speaker N\" (at most 80 characters). An empty string clears it back to \"Speaker N\".",
542
- "examples": [
543
- "Maria Lopez"
544
- ],
545
- "maxLength": 80,
546
- "type": "string"
547
- },
548
- "speaker": {
549
- "description": "The speaker's number: the N of \"Speaker N\" in get_transcription's text, or speakerNumber in its segments (also there once the speaker has a name).",
550
- "examples": [
551
- 2
552
- ],
553
- "minimum": 1,
554
- "type": "integer"
555
- },
556
394
  "transcriptionId": {
557
- "description": "The transcriptionId that transcribe_url, transcribe_file, list_transcriptions, a search tool or start_broadcast returned.",
395
+ "description": "A transcriptionId from list_transcriptions or create_transcription.",
558
396
  "examples": [
559
397
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
560
398
  ],
@@ -563,29 +401,44 @@
563
401
  }
564
402
  },
565
403
  "required": [
566
- "transcriptionId",
567
- "speaker",
568
- "name"
404
+ "transcriptionId"
569
405
  ],
570
406
  "type": "object"
571
407
  },
572
408
  "annotations": {
573
- "title": "Rename a speaker",
409
+ "title": "Delete a transcription",
574
410
  "readOnlyHint": false,
575
- "destructiveHint": false,
411
+ "destructiveHint": true,
576
412
  "idempotentHint": true,
577
413
  "openWorldHint": false
578
414
  }
579
415
  },
580
416
  {
581
- "name": "delete_transcription",
582
- "title": "Delete a transcription",
583
- "description": "Delete a transcription: it moves to the trash with its recordings, transcript, translations and summary, and its share links stop working. It can be restored for 30 days (Dashboard trash, or POST /v1/transcriptions/{id}/restore), then it is erased for good; list_transcriptions with lifecycle \"archived\" lists the trash. A file or link still being transcribed is stopped first. It refuses one that is recording live: stop it first (stop_broadcast for a broadcast). Deleting one already in the trash changes nothing. Returns transcriptionId, name, deleted, deletedAt and restorableUntil. Errors: invalid_state while recording live or when the transcription is locked; not_found for an unknown id; forbidden for a read-only key.",
417
+ "name": "set_cloned_voices",
418
+ "title": "Turn cloned voices on or off",
419
+ "description": "Turn cloned voices on or off for spoken translations: for one transcription, or for one saved speaker everywhere they are recognised.\nOn: the 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. Voices are chosen by VocalStack; this is the only voice choice.\nArguments: give exactly one of transcriptionId (every speaker in it) or speakerId (a saved speaker, from list_speakers). on is required; consent must be true to turn on.\nConsent: turning on records, with your account and the time, that everyone speaking (or that saved speaker) agreed. For a transcription it releases its clones waiting for consent; speakers who refused are never cloned. For a saved speaker saved without consent it starts their clone from their stored sample. Off keeps the saved speaker's fingerprint, consent and clone, only unused; to erase the clone and their samples: delete_speaker.\nBehaviour: free and reversible; lines read aloud after the change use the new setting.\nReturns: transcriptionId, name, clonedVoices; or, with speakerId, speakerId, name, consent, clonedVoice, and voiceState (training, then ready) when cloned.\nErrors: invalid_options for on without consent or both or neither target; invalid_state for a deleted transcription, a speaker who refused, or one with no stored sample to clone from; not_found; forbidden for a read-only key.",
584
420
  "inputSchema": {
585
421
  "additionalProperties": false,
586
422
  "properties": {
423
+ "consent": {
424
+ "default": false,
425
+ "description": "Must be true to turn cloned voices on: confirms everyone speaking (with speakerId: that person) agreed to have their voice cloned. Recorded with your account and the time. Not needed to turn them off.",
426
+ "type": "boolean"
427
+ },
428
+ "on": {
429
+ "description": "true: translations of the speaker's own words are read in their own cloned voice; false (the default for a transcription): stock voices.",
430
+ "type": "boolean"
431
+ },
432
+ "speakerId": {
433
+ "description": "Instead of transcriptionId, for one saved speaker everywhere they are recognised: id in list_speakers, or speakerId from save_speaker.",
434
+ "examples": [
435
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
436
+ ],
437
+ "format": "uuid",
438
+ "type": "string"
439
+ },
587
440
  "transcriptionId": {
588
- "description": "The transcriptionId that transcribe_url, transcribe_file, list_transcriptions, a search tool or start_broadcast returned.",
441
+ "description": "For one transcription's spoken translations: a transcriptionId from create_transcription, list_transcriptions, search_transcriptions or start_broadcast. Leave out with speakerId.",
589
442
  "examples": [
590
443
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
591
444
  ],
@@ -594,41 +447,29 @@
594
447
  }
595
448
  },
596
449
  "required": [
597
- "transcriptionId"
450
+ "on"
598
451
  ],
599
452
  "type": "object"
600
453
  },
601
454
  "annotations": {
602
- "title": "Delete a transcription",
455
+ "title": "Turn cloned voices on or off",
603
456
  "readOnlyHint": false,
604
- "destructiveHint": true,
457
+ "destructiveHint": false,
605
458
  "idempotentHint": true,
606
459
  "openWorldHint": false
607
460
  }
608
461
  },
609
462
  {
610
- "name": "list_voices",
611
- "title": "List voices",
612
- "description": "List the voices that can read translations aloud in one language: stock voices and this account's own cloned voices. For information only: VocalStack picks the voice for each speaker and no tool takes a voice id; the only choice is cloned voices on or off per transcription (set_cloned_voices). For the people the account has saved, use list_speakers. Read-only and free. Returns lang and voices, each with id, name, default (used first), cloned, quality, gender and accent when known, and previewUrl (a short sample). A language with no voices: an empty list.",
463
+ "name": "list_speakers",
464
+ "title": "List saved speakers",
465
+ "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.",
613
466
  "inputSchema": {
614
467
  "additionalProperties": false,
615
- "properties": {
616
- "lang": {
617
- "description": "Code of the language the voices should speak, e.g. en, ro, es.",
618
- "examples": [
619
- "en",
620
- "ro"
621
- ],
622
- "type": "string"
623
- }
624
- },
625
- "required": [
626
- "lang"
627
- ],
468
+ "properties": {},
628
469
  "type": "object"
629
470
  },
630
471
  "annotations": {
631
- "title": "List voices",
472
+ "title": "List saved speakers",
632
473
  "readOnlyHint": true,
633
474
  "destructiveHint": false,
634
475
  "idempotentHint": true,
@@ -636,23 +477,36 @@
636
477
  }
637
478
  },
638
479
  {
639
- "name": "set_cloned_voices",
640
- "title": "Turn cloned voices on or off",
641
- "description": "Turn cloned voices on or off for one transcription's spoken translations. On: each speaker's translated words are read in that speaker's own cloned voice, never anyone else's; a speaker without a usable clone is read in a stock voice. Off (the default): stock voices for everyone. This is the only voice choice; VocalStack picks the voices (list_voices shows them). Turning on needs consent: true, confirming everyone speaking agreed to have their voice cloned; it is recorded with your account and the time, and grants the clones in this transcription that were waiting for consent. Speakers who refused are never cloned. Free; setting the current value again changes nothing. Returns transcriptionId, name and clonedVoices. Errors: invalid_options without consent, invalid_state for a deleted transcription, not_found, forbidden for a read-only key.",
480
+ "name": "save_speaker",
481
+ "title": "Save a speaker",
482
+ "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.\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: rename_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.",
642
483
  "inputSchema": {
643
484
  "additionalProperties": false,
644
485
  "properties": {
645
486
  "consent": {
646
487
  "default": false,
647
- "description": "Required as 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.",
488
+ "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.",
648
489
  "type": "boolean"
649
490
  },
650
- "on": {
651
- "description": "true: read each speaker's translated words in their own cloned voice; false (the default for every transcription): stock voices.",
652
- "type": "boolean"
491
+ "name": {
492
+ "description": "The person's name, shown in this and later transcriptions.",
493
+ "examples": [
494
+ "Maria Lopez"
495
+ ],
496
+ "maxLength": 60,
497
+ "minLength": 1,
498
+ "type": "string"
499
+ },
500
+ "speaker": {
501
+ "description": "The N of \"Speaker N\" in get_transcription's text, or speakerNumber in its segments. Works while a live transcription is still running.",
502
+ "examples": [
503
+ 2
504
+ ],
505
+ "minimum": 1,
506
+ "type": "integer"
653
507
  },
654
508
  "transcriptionId": {
655
- "description": "The transcriptionId that transcribe_url, transcribe_file, list_transcriptions, a search tool or start_broadcast returned.",
509
+ "description": "A transcriptionId from create_transcription, list_transcriptions, search_transcriptions or start_broadcast.",
656
510
  "examples": [
657
511
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
658
512
  ],
@@ -662,31 +516,44 @@
662
516
  },
663
517
  "required": [
664
518
  "transcriptionId",
665
- "on"
519
+ "speaker",
520
+ "name"
666
521
  ],
667
522
  "type": "object"
668
523
  },
669
524
  "annotations": {
670
- "title": "Turn cloned voices on or off",
525
+ "title": "Save a speaker",
671
526
  "readOnlyHint": false,
672
527
  "destructiveHint": false,
673
- "idempotentHint": true,
528
+ "idempotentHint": false,
674
529
  "openWorldHint": false
675
530
  }
676
531
  },
677
532
  {
678
- "name": "list_speakers",
679
- "title": "List saved speakers",
680
- "description": "List the speakers this account has saved: people named once so they are recognised by name in later transcriptions. Use it to see who is known; names within one transcription come from get_transcription and are set with rename_speaker, which saves no one here; list_voices lists the voices for a language. Read-only and free; no arguments. Returns speakers, newest first, each with id, name, gender when known, cloned (their own voice was cloned to read translations of what they say) and, for cloned ones, voiceReady (ready, switched on and consented to). None saved: an empty list.",
533
+ "name": "delete_speaker",
534
+ "title": "Delete a saved speaker",
535
+ "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.\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: rename_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.",
681
536
  "inputSchema": {
682
537
  "additionalProperties": false,
683
- "properties": {},
538
+ "properties": {
539
+ "speakerId": {
540
+ "description": "A saved speaker's id: id in list_speakers, or speakerId from save_speaker.",
541
+ "examples": [
542
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
543
+ ],
544
+ "format": "uuid",
545
+ "type": "string"
546
+ }
547
+ },
548
+ "required": [
549
+ "speakerId"
550
+ ],
684
551
  "type": "object"
685
552
  },
686
553
  "annotations": {
687
- "title": "List saved speakers",
688
- "readOnlyHint": true,
689
- "destructiveHint": false,
554
+ "title": "Delete a saved speaker",
555
+ "readOnlyHint": false,
556
+ "destructiveHint": true,
690
557
  "idempotentHint": true,
691
558
  "openWorldHint": false
692
559
  }
@@ -694,13 +561,21 @@
694
561
  {
695
562
  "name": "list_broadcasts",
696
563
  "title": "List broadcasts",
697
- "description": "List this account's broadcasts, most recently updated first (up to 500, no paging). A broadcast is a public link where listeners follow a live transcription in their own language. Use it to find a broadcastId for get_broadcast, start_broadcast or stop_broadcast, or to see which are on air; get_broadcast lists one broadcast's past transcriptions, which also appear in list_transcriptions. Read-only and free. lifecycle: active (default), archived (the trash; each marked archived) or all. Returns broadcasts, each with broadcastId, name, link, access, onAir, listenersNow, lastRecordedAt once it has recorded, and while on air transcriptionId (read it with get_transcription), state and onAirSince. None: an empty list.",
564
+ "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.",
698
565
  "inputSchema": {
699
566
  "additionalProperties": false,
700
567
  "properties": {
568
+ "broadcastId": {
569
+ "description": "Optional: return only this broadcast, in detail (its recorded transcriptions and stream addresses too). lifecycle is then ignored.",
570
+ "examples": [
571
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
572
+ ],
573
+ "format": "uuid",
574
+ "type": "string"
575
+ },
701
576
  "lifecycle": {
702
577
  "default": "active",
703
- "description": "Which broadcasts to list: active (the default) those in use; archived those in the trash; all both.",
578
+ "description": "Which broadcasts: active those in use, archived those in the trash, all both.",
704
579
  "enum": [
705
580
  "active",
706
581
  "archived",
@@ -718,39 +593,10 @@
718
593
  "openWorldHint": false
719
594
  }
720
595
  },
721
- {
722
- "name": "get_broadcast",
723
- "title": "Get a broadcast",
724
- "description": "Get one broadcast in detail: whether it is on air, its public link, the transcriptions it recorded and the stream addresses it uses. Use it when you have a broadcastId and need more than list_broadcasts gives; read a transcription's text with get_transcription. Read-only and free. Returns broadcastId, name, link, access, onAir, listenersNow, and while on air transcriptionId, state and onAirSince; transcriptions (up to 20, each with transcriptionId, name, recording and lastRecordedAt) and streams (name, url). Unknown id: not_found.",
725
- "inputSchema": {
726
- "additionalProperties": false,
727
- "properties": {
728
- "broadcastId": {
729
- "description": "The broadcastId that list_broadcasts or start_broadcast returned.",
730
- "examples": [
731
- "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
732
- ],
733
- "format": "uuid",
734
- "type": "string"
735
- }
736
- },
737
- "required": [
738
- "broadcastId"
739
- ],
740
- "type": "object"
741
- },
742
- "annotations": {
743
- "title": "Get a broadcast",
744
- "readOnlyHint": true,
745
- "destructiveHint": false,
746
- "idempotentHint": true,
747
- "openWorldHint": false
748
- }
749
- },
750
596
  {
751
597
  "name": "start_broadcast",
752
598
  "title": "Start a broadcast",
753
- "description": "Start live transcription and translation of a live stream (YouTube live, internet radio, HLS .m3u8 or Icecast). Omit broadcastId to create a new broadcast with a new public link; set it to reuse one and keep its link. Use transcribe_url for a recording that has already ended. Billed per second while it runs; listeners are free. It runs until stop_broadcast or the stream ends, so stop it when done. Each call starts a new run (and, without broadcastId, a new broadcast) unless it repeats an idempotencyKey. Returns broadcastId, name, link (anyone who opens it reads, or with spokenTranslations also hears, the stream in their own language), transcriptionId (read the text so far with get_transcription) and state. Errors: invalid_state (with link) when it is already on air, invalid_options for a non-http(s) address, insufficient_credit (with topUpUrl), not_found for an unknown broadcastId.",
599
+ "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.",
754
600
  "inputSchema": {
755
601
  "additionalProperties": false,
756
602
  "properties": {
@@ -763,15 +609,15 @@
763
609
  "type": "string"
764
610
  },
765
611
  "idempotencyKey": {
766
- "description": "Any unique string of visible ASCII. Send the same key with the same arguments when retrying: for 24 h the first result comes back (with replayed true) and nothing is started twice.",
612
+ "description": "Any unique string of visible ASCII. Retrying with the same key and the same arguments within 24 h returns the first result (with replayed true) and nothing is started twice; the same key with other arguments is refused.",
767
613
  "examples": [
768
- "sunday-2026-11-22"
614
+ "meeting-2026-11-19"
769
615
  ],
770
616
  "maxLength": 255,
771
617
  "type": "string"
772
618
  },
773
619
  "name": {
774
- "description": "Name of a new broadcast, shown to listeners. Defaults to the stream's host. Ignored when broadcastId is set.",
620
+ "description": "Name of a new broadcast, shown to listeners. Defaults to the stream's host. Ignored with broadcastId.",
775
621
  "examples": [
776
622
  "Sunday service"
777
623
  ],
@@ -787,11 +633,11 @@
787
633
  },
788
634
  "spokenTranslations": {
789
635
  "default": false,
790
- "description": "Also read each translation aloud, so listeners can hear it in their language as well as read it. Off by default; listening stays free.",
636
+ "description": "Also read each translation aloud, so listeners can hear it in their language as well as read it. Listening stays free.",
791
637
  "type": "boolean"
792
638
  },
793
639
  "streamUrl": {
794
- "description": "http(s) address of a live stream to transcribe and translate as it plays: a YouTube live, an internet radio station, an HLS (.m3u8) or Icecast stream. For a recording that has ended, use transcribe_url.",
640
+ "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.",
795
641
  "examples": [
796
642
  "https://www.youtube.com/watch?v=jfKfPfyJRdk",
797
643
  "https://stream.example.org/radio.m3u8"
@@ -817,12 +663,12 @@
817
663
  {
818
664
  "name": "stop_broadcast",
819
665
  "title": "Stop a broadcast",
820
- "description": "Stop the live transcription a broadcast is running (started with start_broadcast), which also stops its billing. Use it when the event is over. The transcript is kept (get_transcription, translate_transcription, get_summary; delete_transcription removes it) and the link stays, ready for the next start_broadcast with the same broadcastId. Returns the stopped transcription's status (transcriptionId, name, kind \"live\", state \"finishing\" then \"done\", startedAt, endedAt, durationS) with broadcastId and link. Errors: invalid_state when the broadcast is not on air, not_found for an unknown broadcastId.",
666
+ "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 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.",
821
667
  "inputSchema": {
822
668
  "additionalProperties": false,
823
669
  "properties": {
824
670
  "broadcastId": {
825
- "description": "The broadcastId that list_broadcasts or start_broadcast returned.",
671
+ "description": "A broadcastId from list_broadcasts or start_broadcast.",
826
672
  "examples": [
827
673
  "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
828
674
  ],
@@ -842,6 +688,35 @@
842
688
  "idempotentHint": true,
843
689
  "openWorldHint": false
844
690
  }
691
+ },
692
+ {
693
+ "name": "delete_broadcast",
694
+ "title": "Delete a broadcast",
695
+ "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.",
696
+ "inputSchema": {
697
+ "additionalProperties": false,
698
+ "properties": {
699
+ "broadcastId": {
700
+ "description": "A broadcastId from list_broadcasts or start_broadcast, of a broadcast that is not on air.",
701
+ "examples": [
702
+ "0190f0e0-7c1a-7d3e-9b2a-5f4e3d2c1b0a"
703
+ ],
704
+ "format": "uuid",
705
+ "type": "string"
706
+ }
707
+ },
708
+ "required": [
709
+ "broadcastId"
710
+ ],
711
+ "type": "object"
712
+ },
713
+ "annotations": {
714
+ "title": "Delete a broadcast",
715
+ "readOnlyHint": false,
716
+ "destructiveHint": true,
717
+ "idempotentHint": true,
718
+ "openWorldHint": false
719
+ }
845
720
  }
846
721
  ]
847
722
  }