nixamp 0.26.1 → 0.26.3

package/README.md

@@ -537,22 +537,13 @@ uses its direct model; it does not first translate the audio into English.
 
 ### Hear it in your language
 
-The browser's Transcript panel works with movies, shows, sports, courses,
-podcasts, live channels, and rooms. Choose a language and enable **Play
-translated audio**. This is a listener preference: the video and the room's
-shared playback clock keep running. Turn it off to restore the original sound.
-**Start player captions** transcribes other playback in its detected source
-language. Files are played locally; these explicit controls opt into sending
-short audio clips for processing. Ordinary playback uploads no audio.
-
-**Translate another tab** opens the browser's audio-sharing chooser, so a
-watch party or video hosted on another site can be interpreted too. Select a
-browser tab with **Share audio** enabled; **Stop listening** releases sharing.
-Only audio is sent, even though the browser requires a video track to select
-its source. Sharing support depends on the browser and the source's permissions;
-protected media and sources whose audio cannot be captured remain unavailable.
-The control requests suppression of the source tab's local sound. If a browser
-ignores that option, mute the source tab to avoid hearing both languages.
+Use **Translate audio** beside the player's language menu to hear whatever
+Nixamp is playing in your language. One click starts translation; it selects
+your preferred supported language if the menu is still on Original. Turn it off
+to restore the original audio. The video and the room's shared playback clock
+keep running. **Captions** in the Transcript panel enables text-only recognition.
+Ordinary file playback uploads no audio; enabling captions or translated audio
+opts into processing short clips from the playing media.
 
 Native captions use local Whisper Base through Transformers.js. Optional
 speaker-aware audio uses ElevenLabs **Scribe v2** for native transcription with
@@ -565,18 +556,31 @@ also checked before enabling the audio toggle.
 
 A rolling 15-second audio window advances every 5 seconds. Speaker labels are
 reconciled using overlapping timestamps, with different voices assigned to
-separate speakers. Lower/higher pitch suggests a male/female stock voice;
-ambiguous audio uses a default. Each speaker's voice can be changed in the
-panel. Pitch is not gender identity, and a speaker returning after leaving the
-rolling context may receive a new label. Simultaneous speech and noisy crowds
-can still confuse recognition. Native captions never translate to English as
+separate speakers. Voices are picked from the available stock catalogue without
+inferring a person's gender from pitch; each detected speaker gets an unused
+voice until the catalogue is exhausted. **Audio options** folds away optional
+individual overrides. A speaker returning after leaving the rolling context may
+receive a new label. Simultaneous speech and noisy crowds can still confuse
+recognition. Native captions never translate to English as
 an intermediate recognition step.
 
 Processing has one active request and only the latest pending window per
 listener; speech queues and response sizes are bounded. Old transcript history
 is never spoken. Pause, seek, source changes, and disabling the feature cancel
 queued speech; errors restore the original audio. This is a delayed live
-interpreter, not a promise of exact lip sync or background-music separation.
+interpreter, not a promise of exact lip sync.
+
+Translated playback keeps an approximate version of the original background
+sound. FastEnhancer Web's Tiny model estimates speech locally in a dedicated
+browser worker; the player subtracts that estimate from the aligned source in
+each stereo channel and mixes the remainder with translated voices. Background
+processing adds no API calls or provider charges. It stops with translation;
+recognition starts without waiting for it. If the device cannot keep up or load the model, that
+background branch is silenced while translated speech continues. Separation can
+leave some original speech or remove parts of music and crowd noise; disable
+**Keep background sound** under **Audio options** when needed. This uses
+[FastEnhancer Web](https://github.com/ryyr-ry/fastenhancer-web), under the MIT
+license.
 
 The account server needs `ELEVENLABS_API_KEY`; `NIXAMP_DUBBING=off` disables
 this feature. The key stays on the server. Sign-in is required for speaker

package/dist/live-voice.d.ts

@@ -16,6 +16,7 @@ export interface VoiceRequest {
     voice?: string;
     profile?: VoiceProfile;
     channel?: string;
+    speaker?: string;
 }
 export declare class LiveVoice {
     private readonly key;

package/dist/live-voice.js

@@ -209,10 +209,12 @@ export class LiveVoice {
             throw new SpeechError("this language is not supported by Flash voices", 400);
         const voices = await this.voices();
         signal?.throwIfAborted();
-        const gender = ask.profile === "lower" ? "male" : ask.profile === "higher" ? "female" : "neutral";
+        // The browser chooses unique voices for speakers. Legacy callers get a
+        // stable stock voice; pitch is not used to infer gender.
+        const seed = createHash("sha256").update(`${ask.channel ?? ""}|${ask.speaker ?? ""}`).digest().readUInt32BE(0);
         const voice = ask.voice && ask.voice !== "auto"
             ? voices.find(voice => voice.id === ask.voice)
-            : voices.find(voice => voice.gender === gender) ?? voices[0];
+            : voices[seed % voices.length];
         if (!voice)
             throw new SpeechError("choose an available voice", 400);
         const id = createHash("sha256").update(JSON.stringify([LIVE_VOICE_MODEL, voice.id, ask.language, text])).digest("hex");

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "nixamp",
-  "version": "0.26.1",
+  "version": "0.26.3",
   "description": "It really whips the terminal's ass. A Winamp-shaped audio player for your terminal.",
   "license": "MIT",
   "type": "module",

package/src/live-voice.ts

@@ -10,7 +10,7 @@ export const LIVE_VOICE_MODEL = "eleven_flash_v2_5";
 export const LIVE_VOICE_RATE = 16_000;
 export const LIVE_VOICE_LANGUAGES = new Set("en ja zh de hi fr ko pt it es id nl tr fil pl sv bg ro ar cs el fi hr ms sk da ta uk ru hu no vi".split(" "));
 export interface LiveVoiceChoice { id: string; name: string; gender: string; language: string; }
-export interface VoiceRequest { text: string; language: string; voice?: string; profile?: VoiceProfile; channel?: string; }
+export interface VoiceRequest { text: string; language: string; voice?: string; profile?: VoiceProfile; channel?: string; speaker?: string; }
 const HEADERS = { "content-type": "audio/pcm", "cache-control": "no-store", "x-audio-sample-rate": String(LIVE_VOICE_RATE) };
 const budget = (value: number | undefined, fallback: number): number => Number.isFinite(value) && value! >= 0 ? Math.floor(value!) : fallback;
 
@@ -192,10 +192,12 @@ export class LiveVoice {
     if (!LIVE_VOICE_LANGUAGES.has(ask.language)) throw new SpeechError("this language is not supported by Flash voices", 400);
     const voices = await this.voices();
     signal?.throwIfAborted();
-    const gender = ask.profile === "lower" ? "male" : ask.profile === "higher" ? "female" : "neutral";
+    // The browser chooses unique voices for speakers. Legacy callers get a
+    // stable stock voice; pitch is not used to infer gender.
+    const seed = createHash("sha256").update(`${ask.channel ?? ""}|${ask.speaker ?? ""}`).digest().readUInt32BE(0);
     const voice = ask.voice && ask.voice !== "auto"
       ? voices.find(voice => voice.id === ask.voice)
-      : voices.find(voice => voice.gender === gender) ?? voices[0];
+      : voices[seed % voices.length];
     if (!voice) throw new SpeechError("choose an available voice", 400);
     const id = createHash("sha256").update(JSON.stringify([LIVE_VOICE_MODEL, voice.id, ask.language, text])).digest("hex");
     for (const [key, item] of this.cache) if (item.until < this.now()) this.cache.delete(key);