FIX: update settings

Author: Gimanh

packages/lib/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@palabra-ai/translator",
-    "version": "0.0.5",
+    "version": "0.0.6",
     "private": false,
     "main": "dist/lib.js",
     "types": "dist/index.d.ts",
@@ -20,7 +20,8 @@
     },
     "dependencies": {
         "livekit-client": "2.13.0",
-        "typed-emitter": "^2.1.0"
+        "typed-emitter": "^2.1.0",
+        "ts-deepmerge": "^7.0.3"
     },
     "devDependencies": {
         "@eslint/js": "^9.28.0",

packages/lib/src/PalabraClient.model.ts

@@ -1,5 +1,6 @@
 import { SourceLangCode } from '~/utils/source';
 import { TargetLangCode } from '~/utils/target';
+import { PipelineConfigManager } from './config/PipelineConfigManager';
 
 export interface ClientCredentialsAuth {
   clientId: string;
@@ -10,16 +11,19 @@ export interface UserTokenAuth {
   userToken: string;
 }
 
-export interface PalabraClientData {
+
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+export interface PalabraClientData<CM extends PipelineConfigManager<any> = PipelineConfigManager<any>> {
   auth:ClientCredentialsAuth | UserTokenAuth;
   translateFrom: SourceLangCode;
   translateTo: TargetLangCode;
   handleOriginalTrack: () => Promise<MediaStreamTrack>;
-  transportType?: 'webrtc'; // TODO: add websocket transport | 'websocket'
+  transportType?: 'webrtc';
   apiBaseUrl?: string;
   intent?:string;
   audioContext?:AudioContext;
   ignoreAudioContext?:boolean;
+  configManager?: CM;
 }
 
 export type TrackSid = string;

packages/lib/src/PalabraClient.ts

@@ -20,7 +20,8 @@ import { supportsAudioContextSetSinkId, VolumeNode } from './utils';
 import { ConnectionState } from 'livekit-client';
 import { PipelineConfig } from './config';
 
-export class PalabraClient extends PalabraBaseEventEmitter {
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+export class PalabraClient<CM extends PipelineConfigManager<any> = PipelineConfigManager<unknown>> extends PalabraBaseEventEmitter {
   private translateFrom: SourceLangCode;
   private translateTo: TargetLangCode;
   private auth: PalabraClientData['auth'];
@@ -30,7 +31,7 @@ export class PalabraClient extends PalabraBaseEventEmitter {
   private originalTrackVolumeNode: VolumeNode | null = null;
   public transport: PalabraWebRtcTransport | null = null;
   private transportType: PalabraClientData['transportType'];
-  private configManager: PipelineConfigManager;
+  private configManager: CM;
   private audioContext: AudioContext;
 
 
@@ -50,7 +51,7 @@ export class PalabraClient extends PalabraBaseEventEmitter {
 
   private ignoreAudioContext: PalabraClientData['ignoreAudioContext'];
 
-  constructor(data: PalabraClientData) {
+  constructor(data: PalabraClientData<CM>) {
     super();
 
     this.auth = data.auth;
@@ -61,20 +62,21 @@ export class PalabraClient extends PalabraBaseEventEmitter {
 
     this.transportType = data.transportType ?? 'webrtc';
 
-    this.initConfig();
-
     this.shouldPlayTranslation = false;
 
     this.ignoreAudioContext = data.ignoreAudioContext ?? false;
 
-    if (data.audioContext) {
-      this.audioContext = data.audioContext;
+    if (data.configManager) {
+      this.configManager = data.configManager;
     }
+
+    this.initConfig();
+
+    this.initAudioContext(data.audioContext);
   }
 
   public async startTranslation(): Promise<boolean> {
     try {
-      this.initAudioContext();
       await this.wrapOriginalTrack();
       const transport = await this.createSession();
       this.initTransportHandlers();
@@ -93,7 +95,6 @@ export class PalabraClient extends PalabraBaseEventEmitter {
     await this.deleteSession();
     this.transport = null;
     this.stopPlayback();
-    this.closeAudioContext();
     this.cleanUnusedTracks([]);
     this.translationStatus = 'stopped';
     this.cleanupOriginalTrack();
@@ -247,7 +248,7 @@ export class PalabraClient extends PalabraBaseEventEmitter {
     return this.transport;
   }
 
-  public getConfigManager() {
+  public getConfigManager(): CM {
     return this.configManager;
   }
 
@@ -317,6 +318,7 @@ export class PalabraClient extends PalabraBaseEventEmitter {
 
   public async cleanup() {
     await this.stopTranslation();
+    this.closeAudioContext();
     this.initConfig();
   }
 
@@ -337,9 +339,9 @@ export class PalabraClient extends PalabraBaseEventEmitter {
     });
   }
 
-  private async initAudioContext() {
+  private async initAudioContext(audioContext?: AudioContext) {
     if (this.audioContext || this.ignoreAudioContext) return;
-    this.audioContext = new AudioContext();
+    this.audioContext = audioContext ?? new AudioContext();
   }
 
   private closeAudioContext() {
@@ -348,7 +350,9 @@ export class PalabraClient extends PalabraBaseEventEmitter {
   }
 
   private initConfig() {
-    this.configManager = new PipelineConfigManager(this.transportType);
+    if (!this.configManager) {
+      this.configManager = new PipelineConfigManager() as CM;
+    }
 
     this.configManager.setSourceLanguage(this.translateFrom as SourceLangCode);
     this.configManager.addTranslationTarget({ target_language: this.translateTo as TargetLangCode });

packages/lib/src/__tests__/PalabraClient.test.ts

@@ -3,6 +3,7 @@ import { PalabraClient } from '../PalabraClient';
 import type { TargetLangCode } from '../utils/target';
 import type { SourceLangCode } from '../utils/source';
 import { EVENT_START_TRANSLATION, EVENT_STOP_TRANSLATION } from '../transport/PalabraWebRtcTransport.model';
+import { PipelineConfigManager } from '~/config';
 
 // Mock MediaStreamTrack for tests
 class MockMediaStreamTrack {
@@ -140,6 +141,88 @@ describe('PalabraClient', () => {
     client = new PalabraClient(baseConstructorData);
   });
 
+  describe('ConfigManager', () => {
+    it('should set value and get value with extensions', ()=>{
+      const manager = new PipelineConfigManager({ initialExtension: { testProp: 12 } });
+
+      const cl = new PalabraClient({ ...baseConstructorData, configManager: manager });
+      expect((cl.getConfigManager()).getValue('testProp')).toBe(12);
+
+      (cl.getConfigManager()).setValue('testProp', 13);
+      expect((cl.getConfigManager()).getValue('testProp')).toBe(13);
+
+      manager.setValue('testProp', 14);
+      expect((cl.getConfigManager()).getValue('testProp')).toBe(14);
+    });
+
+    it('should set value and get value without extensions', ()=>{
+      const cl = new PalabraClient({ ...baseConstructorData });
+      expect((cl.getConfigManager()).getValue('preprocessing.enable_vad')).toBe(true);
+    });
+  });
+
+  describe('AudioContext', () => {
+    it('should not close AudioContext in stopTranslation', async () => {
+      const closeAudioContextSpy = vi.spyOn(client as unknown as { closeAudioContext: () => void }, 'closeAudioContext').mockImplementation(() => undefined);
+      await client.stopTranslation();
+      expect(closeAudioContextSpy).not.toHaveBeenCalled();
+    });
+
+    it('should not call close method on AudioContext in stopTranslation', async() => {
+      const ctx = new AudioContext();
+      const closeAudioContextSpy = vi.spyOn(ctx, 'close').mockImplementation(() => undefined);
+      const localClient = new PalabraClient({ ...baseConstructorData, audioContext: ctx });
+      await localClient.stopTranslation();
+      expect(closeAudioContextSpy).not.toHaveBeenCalled();
+    });
+
+    it('should close AudioContext in cleanup', async () => {
+      const closeAudioContextSpy = vi.spyOn(client as unknown as { closeAudioContext: () => void }, 'closeAudioContext').mockImplementation(() => undefined);
+      await client.cleanup();
+      expect(closeAudioContextSpy).toHaveBeenCalled();
+    });
+
+    it('should ignore AudioContext creation when ignoreAudioContext is true and audioContext is provided', async() => {
+      const ctx = new AudioContext();
+      const localClient = new PalabraClient({ ...baseConstructorData, audioContext: ctx, ignoreAudioContext: true });
+      // @ts-expect-error: audioContext is private
+      expect(localClient.audioContext).toBeUndefined();
+    });
+
+    it('should create a new AudioContext when ignoreAudioContext is false and audioContext is not provided', async() => {
+      const localClient = new PalabraClient({ ...baseConstructorData, ignoreAudioContext: false });
+      // @ts-expect-error: audioContext is private
+      expect(localClient.audioContext).toBeDefined();
+    });
+
+    it('should use provided AudioContext', async() => {
+      const ctx = new AudioContext();
+      // @ts-expect-error field for test
+      ctx.field = 'test';
+
+      const localClient = new PalabraClient({ ...baseConstructorData, audioContext: ctx, ignoreAudioContext: false });
+      // @ts-expect-error: audioContext is private
+      expect(localClient.audioContext).toBeDefined();
+      // @ts-expect-error: field for test
+      expect(localClient.audioContext.field).toBe('test');
+    });
+
+    it('should create a new AudioContext when audioContext is not provided', async() => {
+      const localClient = new PalabraClient(baseConstructorData);
+      await localClient.startTranslation();
+      // @ts-expect-error: audioContext is private
+      expect(localClient.audioContext).toBeDefined();
+    });
+  });
+
+  it('should get api client', () => {
+    expect(client.getApiClient()).toBeDefined();
+  });
+
+  it('should get config manager', () => {
+    expect(client.getConfigManager()).toBeDefined();
+  });
+
   it('should create a new PalabraClient', () => {
     expect(client).toBeDefined();
   });

packages/lib/src/config/PipelineConfig.model.ts

@@ -6,6 +6,7 @@ export interface StreamConfigBase {
 }
 
 export interface StreamConfigWebRtc extends StreamConfigBase {
+  content_type: 'audio';
   source?: {
     type: 'webrtc';
   };
@@ -43,55 +44,18 @@ export interface PreprocessingConfig {
 
 export interface SentenceSplitterConfig {
   enabled: boolean;
-  splitter_model: 'auto' | string;
-  advanced: {
-    min_sentence_characters: number;
-    min_sentence_seconds: number;
-    min_split_interval: number;
-    context_size: number;
-    segments_after_restart: number;
-    step_size: number;
-    max_steps_without_eos: number;
-    force_end_of_segment: number;
-  };
 }
 
 export interface VerificationConfig {
-  verification_model: 'auto' | string;
-  allow_verification_glossaries: boolean;
   auto_transcription_correction: boolean;
   transcription_correction_style: string | null;
 }
-
-export interface TranscriptionAdvancedConfig {
-  filler_phrases: {
-    enabled: boolean;
-    min_transcription_len: number;
-    min_transcription_time: number;
-    phrase_chance: number;
-  };
-  ignore_languages: SourceLangCode[];
-}
-
 export interface TranscriptionConfig {
   source_language: SourceLangCode;
   detectable_languages: SourceLangCode[];
-  asr_model: 'auto' | string;
-  denoise: 'none' | string;
-  allow_hotwords_glossaries: boolean;
-  supress_numeral_tokens: boolean;
-  diarize_speakers: boolean;
-  priority: 'normal' | string;
-  min_alignment_score: number;
-  max_alignment_cer: number;
   segment_confirmation_silence_threshold: number;
-  only_confirm_by_silence: boolean;
-  batched_inference: boolean;
-  force_detect_language: boolean;
-  calculate_voice_loudness: boolean;
   sentence_splitter: SentenceSplitterConfig;
   verification: VerificationConfig;
-  advanced: TranscriptionAdvancedConfig;
 }
 
 export type AddTranslationArgs = Partial<Omit<TranslationConfig, 'target_language'>> & Pick<TranslationConfig, 'target_language'>;
@@ -102,33 +66,15 @@ export interface VoiceTimbreDetectionConfig {
   low_timbre_voices: string[];
 }
 
-export interface SpeechGenerationAdvancedConfig {
-  f0_variance_factor: number;
-  energy_variance_factor: number;
-  with_custom_stress: boolean;
-}
-
 export interface SpeechGenerationConfig {
-  tts_model: 'auto' | string;
   voice_cloning: boolean;
-  voice_cloning_mode: 'static_10' | string;
-  denoise_voice_samples: boolean;
   voice_id: string;
   voice_timbre_detection: VoiceTimbreDetectionConfig;
-  speech_tempo_auto: boolean;
-  speech_tempo_timings_factor: number;
-  speech_tempo_adjustment_factor: number;
-  advanced: SpeechGenerationAdvancedConfig;
 }
 
 export interface TranslationConfig {
   target_language: TargetLangCode;
-  allowed_source_languages: SourceLangCode[];
-  translation_model: 'auto' | 'alpha' | string;
-  allow_translation_glossaries: boolean;
-  style: string | null;
   translate_partial_transcriptions: boolean;
-  advanced: Record<string, unknown>;
   speech_generation: SpeechGenerationConfig;
 }
 
@@ -146,8 +92,7 @@ export type AllowedMessageTypes = (string
   | 'translated_transcription'
   | 'partial_translated_transcription'
   | 'partial_transcription'
-  | 'validated_transcription'
-  | 'pipeline_timings');
+  | 'validated_transcription');
 
 export interface PipelineConfig {
   input_stream: StreamConfig;