Added hugging face token

package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@ztimson/ai-utils",
-	"version": "0.6.9",
+	"version": "0.6.10",
 	"description": "AI Utility library",
 	"author": "Zak Timson",
 	"license": "MIT",

src/ai.ts

@@ -8,6 +8,8 @@ export type AbortablePromise<T> = Promise<T> & {
 };
 
 export type AiOptions = {
+	/** Token to pull models from hugging face */
+	hfToken?: string;
 	/** Path to models */
 	path?: string;
 	/** ASR model: whisper-tiny, whisper-base */

src/asr.ts

@@ -14,14 +14,15 @@ export async function canDiarization(): Promise<boolean> {
 	});
 }
 
-async function runDiarization(audioPath: string, torchHome: string): Promise<any[]> {
+async function runDiarization(audioPath: string, dir: string, token: string): Promise<any[]> {
 	const script = `
 import sys
 import json
 import os
 from pyannote.audio import Pipeline
 
-os.environ['TORCH_HOME'] = "${torchHome}"
+os.environ['TORCH_HOME'] = "${dir}"
+os.environ['HF_TOKEN'] = "${token}"
 pipeline = Pipeline.from_pretrained("pyannote/speaker-diarization-3.1")
 diarization = pipeline(sys.argv[1])
 
@@ -82,7 +83,7 @@ function combineSpeakerTranscript(chunks: any[], speakers: any[]): string {
 	return lines.join('\n');
 }
 
-parentPort?.on('message', async ({ file, speaker, model, modelDir }) => {
+parentPort?.on('message', async ({ file, speaker, model, modelDir, token }) => {
 	try {
 		console.log('worker', file);
 		if(!whisperPipeline) whisperPipeline = await pipeline('automatic-speech-recognition', `Xenova/${model}`, {cache_dir: modelDir, quantized: true});
@@ -111,12 +112,12 @@ parentPort?.on('message', async ({ file, speaker, model, modelDir }) => {
 
 		// Speaker Diarization
 		const hasDiarization = await canDiarization();
-		if(!hasDiarization) {
+		if(!token || !hasDiarization) {
 			parentPort?.postMessage({ text: transcriptResult.text?.trim() || null, error: 'Speaker diarization unavailable' });
 			return;
 		}
 
-		const speakers = await runDiarization(file, modelDir);
+		const speakers = await runDiarization(file, modelDir, token);
 		const combined = combineSpeakerTranscript(transcriptResult.chunks || [], speakers);
 		parentPort?.postMessage({ text: combined });
 	} catch (err) {

src/audio.ts

@@ -32,7 +32,7 @@ export class Audio {
 			worker.on('exit', (code) => {
 				if(code !== 0 && !aborted) reject(new Error(`Worker exited with code ${code}`));
 			});
-			worker.postMessage({file, model, speaker, modelDir: this.ai.options.path});
+			worker.postMessage({file, model, speaker, modelDir: this.ai.options.path, token: this.ai.options.hfToken});
 		});
 		return Object.assign(p, { abort });
 	}