opc-agent 4.1.0 → 4.1.2

package/src/core/audio.ts

@@ -1,98 +1,98 @@
-/**
- * Audio Processor - v1.0.0
- * Audio transcription/synthesis wrappers with format detection, duration, split.
- */
-
-export type AudioFormat = 'wav' | 'mp3' | 'ogg' | 'flac' | 'webm' | 'aac' | 'unknown';
-
-export interface TranscribeOptions {
-  language?: string;
-  model?: string;
-  provider?: string;
-}
-
-export interface SynthesizeOptions {
-  voice?: string;
-  speed?: number;
-  format?: AudioFormat;
-  provider?: string;
-}
-
-export interface TranscribeResult {
-  text: string;
-  language?: string;
-  duration?: number;
-  segments?: { start: number; end: number; text: string }[];
-}
-
-export interface SynthesizeResult {
-  audio: Buffer;
-  format: AudioFormat;
-  duration?: number;
-}
-
-export type TranscribeFunction = (audio: Buffer, options?: TranscribeOptions) => Promise<TranscribeResult>;
-export type SynthesizeFunction = (text: string, options?: SynthesizeOptions) => Promise<SynthesizeResult>;
-
-const FORMAT_SIGNATURES: [Buffer, AudioFormat][] = [
-  [Buffer.from('RIFF'), 'wav'],
-  [Buffer.from([0xff, 0xfb]), 'mp3'],
-  [Buffer.from([0xff, 0xf3]), 'mp3'],
-  [Buffer.from([0xff, 0xf2]), 'mp3'],
-  [Buffer.from([0x49, 0x44, 0x33]), 'mp3'], // ID3
-  [Buffer.from('OggS'), 'ogg'],
-  [Buffer.from('fLaC'), 'flac'],
-  [Buffer.from([0x1a, 0x45, 0xdf, 0xa3]), 'webm'],
-];
-
-export class AudioProcessor {
-  private transcribeFn?: TranscribeFunction;
-  private synthesizeFn?: SynthesizeFunction;
-
-  constructor(options?: { transcribe?: TranscribeFunction; synthesize?: SynthesizeFunction }) {
-    this.transcribeFn = options?.transcribe;
-    this.synthesizeFn = options?.synthesize;
-  }
-
-  /** Detect audio format from buffer header */
-  static detectFormat(audio: Buffer): AudioFormat {
-    for (const [sig, fmt] of FORMAT_SIGNATURES) {
-      if (audio.length >= sig.length && audio.subarray(0, sig.length).equals(sig)) return fmt;
-    }
-    return 'unknown';
-  }
-
-  /** Estimate duration in seconds for WAV files (for others returns undefined) */
-  static estimateDuration(audio: Buffer): number | undefined {
-    const fmt = AudioProcessor.detectFormat(audio);
-    if (fmt === 'wav' && audio.length >= 44) {
-      const sampleRate = audio.readUInt32LE(24);
-      const byteRate = audio.readUInt32LE(28);
-      if (byteRate > 0) {
-        const dataSize = audio.length - 44;
-        return dataSize / byteRate;
-      }
-    }
-    return undefined;
-  }
-
-  /** Split audio buffer into chunks of roughly `chunkBytes` size */
-  static split(audio: Buffer, chunkBytes: number): Buffer[] {
-    if (chunkBytes <= 0) throw new Error('chunkBytes must be positive');
-    const chunks: Buffer[] = [];
-    for (let i = 0; i < audio.length; i += chunkBytes) {
-      chunks.push(audio.subarray(i, Math.min(i + chunkBytes, audio.length)));
-    }
-    return chunks;
-  }
-
-  async transcribe(audio: Buffer, options?: TranscribeOptions): Promise<TranscribeResult> {
-    if (!this.transcribeFn) throw new Error('No transcribe provider configured');
-    return this.transcribeFn(audio, options);
-  }
-
-  async synthesize(text: string, options?: SynthesizeOptions): Promise<SynthesizeResult> {
-    if (!this.synthesizeFn) throw new Error('No synthesize provider configured');
-    return this.synthesizeFn(text, options);
-  }
-}
+/**
+ * Audio Processor - v1.0.0
+ * Audio transcription/synthesis wrappers with format detection, duration, split.
+ */
+
+export type AudioFormat = 'wav' | 'mp3' | 'ogg' | 'flac' | 'webm' | 'aac' | 'unknown';
+
+export interface TranscribeOptions {
+  language?: string;
+  model?: string;
+  provider?: string;
+}
+
+export interface SynthesizeOptions {
+  voice?: string;
+  speed?: number;
+  format?: AudioFormat;
+  provider?: string;
+}
+
+export interface TranscribeResult {
+  text: string;
+  language?: string;
+  duration?: number;
+  segments?: { start: number; end: number; text: string }[];
+}
+
+export interface SynthesizeResult {
+  audio: Buffer;
+  format: AudioFormat;
+  duration?: number;
+}
+
+export type TranscribeFunction = (audio: Buffer, options?: TranscribeOptions) => Promise<TranscribeResult>;
+export type SynthesizeFunction = (text: string, options?: SynthesizeOptions) => Promise<SynthesizeResult>;
+
+const FORMAT_SIGNATURES: [Buffer, AudioFormat][] = [
+  [Buffer.from('RIFF'), 'wav'],
+  [Buffer.from([0xff, 0xfb]), 'mp3'],
+  [Buffer.from([0xff, 0xf3]), 'mp3'],
+  [Buffer.from([0xff, 0xf2]), 'mp3'],
+  [Buffer.from([0x49, 0x44, 0x33]), 'mp3'], // ID3
+  [Buffer.from('OggS'), 'ogg'],
+  [Buffer.from('fLaC'), 'flac'],
+  [Buffer.from([0x1a, 0x45, 0xdf, 0xa3]), 'webm'],
+];
+
+export class AudioProcessor {
+  private transcribeFn?: TranscribeFunction;
+  private synthesizeFn?: SynthesizeFunction;
+
+  constructor(options?: { transcribe?: TranscribeFunction; synthesize?: SynthesizeFunction }) {
+    this.transcribeFn = options?.transcribe;
+    this.synthesizeFn = options?.synthesize;
+  }
+
+  /** Detect audio format from buffer header */
+  static detectFormat(audio: Buffer): AudioFormat {
+    for (const [sig, fmt] of FORMAT_SIGNATURES) {
+      if (audio.length >= sig.length && audio.subarray(0, sig.length).equals(sig)) return fmt;
+    }
+    return 'unknown';
+  }
+
+  /** Estimate duration in seconds for WAV files (for others returns undefined) */
+  static estimateDuration(audio: Buffer): number | undefined {
+    const fmt = AudioProcessor.detectFormat(audio);
+    if (fmt === 'wav' && audio.length >= 44) {
+      const sampleRate = audio.readUInt32LE(24);
+      const byteRate = audio.readUInt32LE(28);
+      if (byteRate > 0) {
+        const dataSize = audio.length - 44;
+        return dataSize / byteRate;
+      }
+    }
+    return undefined;
+  }
+
+  /** Split audio buffer into chunks of roughly `chunkBytes` size */
+  static split(audio: Buffer, chunkBytes: number): Buffer[] {
+    if (chunkBytes <= 0) throw new Error('chunkBytes must be positive');
+    const chunks: Buffer[] = [];
+    for (let i = 0; i < audio.length; i += chunkBytes) {
+      chunks.push(audio.subarray(i, Math.min(i + chunkBytes, audio.length)));
+    }
+    return chunks;
+  }
+
+  async transcribe(audio: Buffer, options?: TranscribeOptions): Promise<TranscribeResult> {
+    if (!this.transcribeFn) throw new Error('No transcribe provider configured');
+    return this.transcribeFn(audio, options);
+  }
+
+  async synthesize(text: string, options?: SynthesizeOptions): Promise<SynthesizeResult> {
+    if (!this.synthesizeFn) throw new Error('No synthesize provider configured');
+    return this.synthesizeFn(text, options);
+  }
+}