friend-assistant 0.1.43 → 0.1.44

package/README.md

@@ -138,7 +138,7 @@ Connectors (Gmail, Chrome control, any MCP service) are managed inside the app,
 | `ollama.url`, `ollama.model`, `ollama.vision`, `ollama.contextTokens` | The local model: server address, model name, whether it can see pictures, and how much it remembers at once. |
 | `openai.url`, `openai.model`, `openai.apiKey`, `openai.vision` | An OpenAI-compatible server: its address (ending in `/v1`), the model name it knows, its API key (or empty), and whether the model can see pictures. For OmniRoute (https://github.com/diegosouzapw/OmniRoute) run `npm install -g omniroute` and `omniroute`, then `friend config set provider openai` (the defaults are OmniRoute's: `http://localhost:20128/v1` and model `auto`). Also works with LM Studio, LiteLLM, vLLM and OpenRouter. |
 | `voice.hindiVoice`, `voice.marathiVoice` | The voices that read Hindi and Marathi (Devanagari) text: `hi_IN-priyamvada-medium`, `hi_IN-pratham-medium`, `hi_IN-rohan-medium`, `mr_IN-google-medium`. Empty makes the main voice read that language. See "Hindi, Marathi and English together". |
-| `voice.enabled`, `voice.sttModel`, `voice.ttsVoice`, `voice.language`, `voice.speakTyped`, `voice.accept`, `voice.challenge` | Speech on or off, the listening model, the speaking voice, a forced language, speaking typed replies, and the two thresholds for recognising your voice. |
+| `voice.enabled`, `voice.sttModel`, `voice.ttsVoice`, `voice.language`, `voice.speakTyped`, `voice.duck`, `voice.duckPercent`, `voice.accept`, `voice.challenge` | Speech on or off, the listening model, the speaking voice, a forced language, speaking typed replies, turning music and video down while you talk and while Friend speaks (and how far, in percent), and the two thresholds for recognising your voice. |
 | `bind`, `port` | Where Friend listens (easier: `friend lan`). |
 | `https` | `true` serves a secure `https://` address with a certificate Friend makes for itself, so a phone may use the microphone (easier: `friend lan https on`). |
 | `publicUrl`, `allowedOrigins`, `allowedHosts`, `trustProxy` | For running behind your own domain or reverse proxy (see "Using your own domain later"). |

package/dist/BUILD.json

@@ -1,5 +1,5 @@
 {
   "name": "friend-assistant",
-  "version": "0.1.43",
-  "builtAt": "2026-10-13T05:55:13.217Z"
+  "version": "0.1.44",
+  "builtAt": "2026-10-13T06:25:53.644Z"
 }

package/dist/friend.mjs

@@ -312,6 +312,8 @@ var ServerMsg = z3.discriminatedUnion("type", [
     data: z3.string()
   }),
   z3.object({ type: z3.literal("audio.stop") }),
+  /** How loud music and video should be while the owner talks or Friend speaks: 1 is as it is, 0.25 a quarter. */
+  z3.object({ type: z3.literal("audio.duck"), level: z3.number().min(0).max(1) }),
   /** The owner approved a look through the camera: take one picture and upload it. */
   z3.object({ type: z3.literal("camera.request"), id: z3.string() }),
   /** Stop: the request was answered elsewhere, timed out, or the emergency lock fired. */
@@ -335,6 +337,8 @@ var ClientMsg = z3.discriminatedUnion("type", [
   z3.object({ type: z3.literal("stage.event"), objectId: z3.string(), event: z3.string().max(64) }),
   ApprovalRespond,
   z3.object({ type: z3.literal("voice.interrupt") }),
+  /** This screen is listening to the owner, or Friend is speaking on it (true), or neither (false). */
+  z3.object({ type: z3.literal("audio.activity"), active: z3.boolean() }),
   z3.object({ type: z3.literal("voice.confirm"), id: z3.string(), pin: z3.string().max(32) }),
   z3.object({ type: z3.literal("voice.cancel"), id: z3.string() }),
   z3.object({ type: z3.literal("camera.denied"), id: z3.string(), reason: z3.enum(["off", "no-camera", "permission", "busy"]) }),
@@ -3352,12 +3356,16 @@ var DEFAULT_VOICE = {
   accept: 0.72,
   challenge: 0.58,
   hindiVoice: DEFAULT_HINDI_VOICE,
-  marathiVoice: DEFAULT_MARATHI_VOICE
+  marathiVoice: DEFAULT_MARATHI_VOICE,
+  duck: true,
+  duckPercent: 25
 };
 function sanitizeVoice(v) {
   const ok = (n) => typeof n === "number" && Number.isFinite(n) && n >= 0 && n <= 1;
-  if (ok(v.accept) && ok(v.challenge) && v.accept > v.challenge) return v;
-  return { ...v, accept: DEFAULT_VOICE.accept, challenge: DEFAULT_VOICE.challenge };
+  const percent2 = typeof v.duckPercent === "number" && Number.isFinite(v.duckPercent) ? Math.max(0, Math.min(100, Math.round(v.duckPercent))) : DEFAULT_VOICE.duckPercent;
+  const base = { ...v, duck: typeof v.duck === "boolean" ? v.duck : DEFAULT_VOICE.duck, duckPercent: percent2 };
+  if (ok(v.accept) && ok(v.challenge) && v.accept > v.challenge) return base;
+  return { ...base, accept: DEFAULT_VOICE.accept, challenge: DEFAULT_VOICE.challenge };
 }
 var DEFAULT_DESTRUCTIVE_PATTERNS = [
   String.raw`\brm\s+(-[a-z]*r|-[a-z]*f)`,
@@ -5292,8 +5300,8 @@ var Mpv = class {
   seek(seconds) {
     return this.cmd(["seek", seconds, "absolute"]);
   }
-  volume(percent) {
-    return this.cmd(["set_property", "volume", percent]);
+  volume(percent2) {
+    return this.cmd(["set_property", "volume", percent2]);
   }
   mute(on) {
     return this.cmd(["set_property", "mute", on]);
@@ -5384,8 +5392,8 @@ var VideoHost = class {
   seek(seconds) {
     return this.mpv.seek(seconds);
   }
-  volume(percent) {
-    return this.mpv.volume(percent);
+  volume(percent2) {
+    return this.mpv.volume(percent2);
   }
   stop() {
     return this.mpv.stop();
@@ -5989,10 +5997,10 @@ var Provisioner = class {
       const list2 = await this.text(plan.sums.url);
       const want = checksumFor(list2, plan.sums.name);
       if (!want) return fail(`Friend could not check the ${name} download: no published checksum was found for it.`);
-      let got = await this.file(plan.url, part, (percent) => this.set(name, { phase: "downloading", percent }));
+      let got = await this.file(plan.url, part, (percent2) => this.set(name, { phase: "downloading", percent: percent2 }));
       if (got.error === "failed") {
         rmSync4(part, { force: true });
-        got = await this.file(plan.url, part, (percent) => this.set(name, { phase: "downloading", percent }));
+        got = await this.file(plan.url, part, (percent2) => this.set(name, { phase: "downloading", percent: percent2 }));
       }
       if (got.error) {
         this.d.onError?.(name, got.error === "big" ? "too big" : got.why ?? "download failed");
@@ -6083,10 +6091,10 @@ var Provisioner = class {
         }
         hash.update(value);
         if (!out.write(value)) await new Promise((r) => out.once("drain", () => r()));
-        const percent = total > 0 ? Math.min(99, Math.floor(received / total * 100)) : null;
-        if (percent !== lastPercent) {
-          lastPercent = percent ?? -1;
-          progress(percent);
+        const percent2 = total > 0 ? Math.min(99, Math.floor(received / total * 100)) : null;
+        if (percent2 !== lastPercent) {
+          lastPercent = percent2 ?? -1;
+          progress(percent2);
         }
       }
       await new Promise((resolve13, reject) => {
@@ -6346,6 +6354,10 @@ var VideoHub = class {
   tokens = /* @__PURE__ */ new Map();
   streams = [];
   cached = null;
+  /** The volume the owner chose for the PC's player (mpv keeps it from one video to the next). */
+  hostVolume = 100;
+  /** How loud it is while someone talks: 1 is the owner's volume as it is. */
+  duckLevel = 1;
   now = null;
   tickets = /* @__PURE__ */ new Map();
   clock() {
@@ -6490,6 +6502,8 @@ var VideoHub = class {
       const opened = await this.d.host.open(info.webpage || args.url || "", { start: start2, maxHeight: cfg.maxHeight });
       if (!opened.ok) return { ok: false, text: opened.error };
       this.now = { output, title: info.title, uploader: info.uploader, info, via: "mpv" };
+      if (this.duckLevel < 1) await this.d.host.volume(this.hostLevel()).catch(() => {
+      });
       return { ok: true, text: `Playing ${what} in a window on this PC.` };
     }
     if (browserWindow) {
@@ -6534,6 +6548,17 @@ var VideoHub = class {
     });
     this.now = null;
   }
+  hostLevel() {
+    return Math.round(this.hostVolume * this.duckLevel);
+  }
+  /** Turns the video in the PC's own player down (0.25 is a quarter) while someone talks, and back up (1) after. A video on the screen is turned down by the page. */
+  async duck(level) {
+    this.duckLevel = Math.max(0, Math.min(1, level));
+    const now = this.now;
+    if (now?.output !== "host" || now.via !== "mpv" || !this.d.host) return;
+    await this.d.host.volume(this.hostLevel()).catch(() => {
+    });
+  }
   async control(args) {
     if (args.action === "stop") {
       const wasScreen = this.now?.output === "screen";
@@ -6549,7 +6574,10 @@ var VideoHub = class {
         if (args.action === "pause") await host.pause(true);
         else if (args.action === "resume") await host.pause(false);
         else if (args.action === "seek") await host.seek(Math.max(0, args.seconds ?? 0));
-        else await host.volume(Math.max(0, Math.min(100, args.level ?? 50)));
+        else {
+          this.hostVolume = Math.max(0, Math.min(100, args.level ?? 50));
+          await host.volume(this.hostLevel());
+        }
       } catch (e) {
         return { ok: false, text: e instanceof Error ? e.message : "The player did not answer." };
       }
@@ -8907,6 +8935,8 @@ var HostPlayer = class {
   chain = Promise.resolve();
   off;
   disposed = false;
+  /** How loud it is while someone talks: 1 is the owner's volume as it is. */
+  duckLevel = 1;
   now = () => (this.d.clock ?? Date.now)();
   available() {
     const problem = programProblem(this.d.settings().hostPlayer.trim());
@@ -9090,14 +9120,24 @@ var HostPlayer = class {
     if (this.engine === "folder") await this.d.mpv.seek(seconds);
     else if (this.engine === "spotify" && this.deviceId) await this.d.spotify.control(this.deviceId, { action: "seek", ms: Math.round(seconds * 1e3) });
   }
+  /** The owner's volume, turned down while someone is talking. */
+  level() {
+    return Math.round(this.s.volume * this.duckLevel);
+  }
   async applyVolume() {
     if (this.engine === "folder") {
-      await this.d.mpv.volume(this.s.volume);
+      await this.d.mpv.volume(this.level());
       await this.d.mpv.mute(this.s.muted);
     } else if (this.engine === "spotify" && this.deviceId) {
-      await this.d.spotify.control(this.deviceId, { action: "volume", percent: this.s.muted ? 0 : this.s.volume });
+      await this.d.spotify.control(this.deviceId, { action: "volume", percent: this.s.muted ? 0 : this.level() });
     }
   }
+  /** Turns the sound down (0.25 is a quarter) while the owner talks to Friend or Friend speaks, and back up (1) after. The owner's own volume is not changed. */
+  async duck(level) {
+    this.duckLevel = Math.max(0, Math.min(1, level));
+    if (this.disposed || !this.engine) return;
+    await this.run(() => this.applyVolume());
+  }
   // ---- what the engines tell us ----
   onMpv(e) {
     if (this.disposed) return;
@@ -12956,7 +12996,9 @@ var SettingsPatch = z11.object({
     /** A Piper voice name; empty makes the main voice read Hindi too. */
     hindiVoice: z11.string().trim().max(80).regex(/^[A-Za-z0-9_-]*$/, "A Piper voice name such as hi_IN-priyamvada-medium, or empty"),
     marathiVoice: z11.string().trim().max(80).regex(/^[A-Za-z0-9_-]*$/, "A Piper voice name such as mr_IN-google-medium, or empty"),
-    sttModel: z11.string().trim().min(1).max(80).regex(/^[\w./-]+$/)
+    sttModel: z11.string().trim().min(1).max(80).regex(/^[\w./-]+$/),
+    duck: z11.boolean(),
+    duckPercent: z11.number().int().min(0).max(100)
   }).partial().strict()
 }).partial().strict();
 function settingsView(c) {
@@ -13017,7 +13059,9 @@ function settingsView(c) {
       ttsVoice: c.voice.ttsVoice,
       sttModel: c.voice.sttModel,
       hindiVoice: c.voice.hindiVoice,
-      marathiVoice: c.voice.marathiVoice
+      marathiVoice: c.voice.marathiVoice,
+      duck: c.voice.duck,
+      duckPercent: c.voice.duckPercent
     }
   };
 }
@@ -13609,6 +13653,42 @@ async function ensureTlsCert(dataDir, extraHosts = [], ifaces, now = Date.now) {
   return { key: pems.private, cert: pems.cert };
 }
 
+// ../server/src/audio/duck.ts
+var FALLBACK_PERCENT = 25;
+var DuckController = class {
+  constructor(d) {
+    this.d = d;
+  }
+  d;
+  active = /* @__PURE__ */ new Set();
+  current = 1;
+  /** How loud things should be now: 1 as normal. */
+  get level() {
+    return this.current;
+  }
+  set(source, active) {
+    if (active) this.active.add(source);
+    else this.active.delete(source);
+    this.update();
+  }
+  /** A screen went away. */
+  drop(source) {
+    if (this.active.delete(source)) this.update();
+  }
+  /** The owner changed the settings. */
+  refresh() {
+    this.update();
+  }
+  update() {
+    const raw = this.d.percent();
+    const percent2 = Number.isFinite(raw) ? Math.max(0, Math.min(100, raw)) : FALLBACK_PERCENT;
+    const next = this.d.enabled() && this.active.size > 0 ? percent2 / 100 : 1;
+    if (next === this.current) return;
+    this.current = next;
+    this.d.apply(next);
+  }
+};
+
 // ../server/src/voice/challenge.ts
 import { randomUUID as randomUUID6 } from "node:crypto";
 var Challenges = class {
@@ -14484,6 +14564,7 @@ async function createApp(opts) {
       else swap();
     }
     if (applied.changed.includes("voice.ttsVoice")) void warmVoice(config.voice.ttsVoice, voiceBefore);
+    if (applied.changed.some((k) => k === "voice.duck" || k === "voice.duckPercent")) ducker.refresh();
     if (applied.changed.includes("voice.sttModel")) void warmListening(config.voice.sttModel, sttBefore);
     if (applied.changed.includes("voice.hindiVoice")) void warmLanguageVoice(config.voice.hindiVoice);
     if (applied.changed.includes("voice.marathiVoice")) void warmLanguageVoice(config.voice.marathiVoice);
@@ -14925,6 +15006,20 @@ async function createApp(opts) {
       return key;
     }
   };
+  const duckIds = /* @__PURE__ */ new WeakMap();
+  let duckSeq = 0;
+  const duckId = (conn) => duckIds.get(conn) ?? (duckIds.set(conn, `screen-${++duckSeq}`), duckIds.get(conn));
+  const ducker = new DuckController({
+    enabled: () => config.voice.duck,
+    percent: () => config.voice.duckPercent,
+    apply: (level) => {
+      broadcast({ type: "audio.duck", level });
+      void hostPlayer.duck(level).catch(() => {
+      });
+      void videoHub.duck(level).catch(() => {
+      });
+    }
+  });
   const tls = config.https ? await ensureTlsCert(opts.dataDir, [...config.allowedHosts, ...config.publicUrl ? [new URL(config.publicUrl).hostname] : []]) : void 0;
   const gateway = createGateway({
     ...tls ? { tls } : {},
@@ -15338,11 +15433,13 @@ ${url}`);
       conn.send(statusMsg());
       conn.send(uiMsg());
       conn.send({ type: "music.host", state: hostPlayer.state() });
+      if (ducker.level < 1) conn.send({ type: "audio.duck", level: ducker.level });
       const pending = challenges.current();
       if (pending) conn.send({ type: "voice.challenge", id: pending.id, text: pending.text });
       return () => {
         clients.delete(conn);
         musicHub.detach(conn);
+        ducker.drop(duckId(conn));
       };
     },
     onClientMessage: (_session, msg, conn) => {
@@ -15360,6 +15457,9 @@ ${url}`);
         case "voice.interrupt":
           speaker?.interrupt(broadcast);
           break;
+        case "audio.activity":
+          ducker.set(duckId(conn), msg.active);
+          break;
         case "voice.confirm":
           void confirmChallenge(msg.id, msg.pin, conn).catch((e) => audit.log("system", "voice.confirm_error", { message: e.message }));
           break;
@@ -15658,6 +15758,11 @@ var unit2 = (v) => {
   if (v.trim() === "" || !Number.isFinite(n) || n < 0 || n > 1) throw new Error("Expected a number between 0 and 1");
   return n;
 };
+var percent = (v) => {
+  const n = Number(v);
+  if (v.trim() === "" || !Number.isFinite(n) || n < 0 || n > 100) throw new Error("Expected a number between 0 and 100");
+  return Math.round(n);
+};
 var httpUrl = (v) => {
   let u;
   try {
@@ -15716,6 +15821,8 @@ var GROUPS = {
     },
     language: (v) => v.trim() === "" ? null : v.trim(),
     speakTyped: bool,
+    duck: bool,
+    duckPercent: percent,
     accept: unit2,
     challenge: unit2
   },
@@ -15780,6 +15887,8 @@ var KEY_HELP = {
   "voice.hindiVoice": "the voice that reads Hindi (Devanagari) text: hi_IN-priyamvada-medium, hi_IN-pratham-medium or hi_IN-rohan-medium; empty uses the main voice",
   "voice.marathiVoice": "the voice that reads Marathi (Devanagari) text: mr_IN-google-medium; empty uses the main voice",
   "voice.speakTyped": "true or false: also speak the replies to typed messages",
+  "voice.duck": "true or false: turn music and video down while you talk to Friend and while it speaks",
+  "voice.duckPercent": "0 to 100: how loud music and video are then, as a percent of how loud they were (25 is a quarter)",
   "voice.accept": "0 to 1: a voice at or above this match is you",
   "voice.challenge": "0 to 1: a voice between this and accept must be confirmed with your PIN",
   "ollama.url": "address of the Ollama server, for example http://127.0.0.1:11434",
@@ -17943,7 +18052,7 @@ async function daemonCommand(args, io2, deps = realDeps) {
 import crossSpawn5 from "cross-spawn";
 var PACKAGE = "friend-assistant";
 var real = {
-  current: () => true ? "0.1.43" : "development",
+  current: () => true ? "0.1.44" : "development",
   latest: () => new Promise((resolve13) => {
     const child = crossSpawn5("npm", ["view", PACKAGE, "version", "--prefer-online"], { stdio: ["ignore", "pipe", "ignore"] });
     let out = "";
@@ -18231,7 +18340,7 @@ async function run(argv, io2) {
       case "version":
       case "--version":
       case "-v":
-        io2.out(true ? "0.1.43" : "development");
+        io2.out(true ? "0.1.44" : "development");
         return 0;
       case "lan":
         return await lanCommand(rest, io2);