STT/TSS exposed in API

Author: 0xGingi

api-docs.md

@@ -104,7 +104,7 @@ curl -X POST "http://localhost:3432/api/generate-message" \
 #### POST `/api/tts`
 Proxies requests to NanoGPT TTS API for speech synthesis.
 
-**Authentication**: Session
+**Authentication**: Session or API Key
 
 **Request Body**:
 ```json
@@ -129,18 +129,48 @@ curl -X POST "http://localhost:3432/api/tts" \
   --output speech.mp3
 ```
 
+#### GET `/api/tts/status`
+Polls the status of an asynchronous TTS request.
+
+**Authentication**: Session or API Key
+
+**Query Parameters**:
+- `runId`: (Required) The run ID returned by `/api/tts`.
+- `model`: (Required) The TTS model ID.
+- `cost`: (Optional) Estimated cost from the submit response.
+- `paymentSource`: (Optional) Payment source from the submit response.
+- `isApiRequest`: (Optional) Whether the request was an API request.
+
+**Response**:
+```json
+{
+  "status": "pending" | "completed" | "error",
+  "audioUrl": "string (when completed)",
+  "contentType": "string (optional)",
+  "model": "string",
+  "error": "string (optional)"
+}
+```
+
+**CURL Example**:
+```bash
+curl "http://localhost:3432/api/tts/status?runId=RUN_ID&model=Elevenlabs-Turbo-V2.5" \
+  -H "Authorization: Bearer your_api_key"
+```
+
 ---
 
 ### Speech-to-Text
 
 #### POST `/api/stt`
 Proxies audio files to NanoGPT STT API for transcription.
 
-**Authentication**: Session
+**Authentication**: Session or API Key
 
 **Request Body** (FormData):
 - `audio`: Binary audio file (webm, mp4, etc).
 - `model`: "string (optional, default: 'Whisper-Large-V3')"
+- `language`: "string (optional, default: 'auto')"
 
 **Response**:
 ```json

src/routes/api/stt/+server.ts

@@ -4,14 +4,26 @@ import type { RequestHandler } from './$types';
 import { db } from '$lib/db';
 import { modelPerformanceStats } from '$lib/db/schema';
 import { eq, and, sql } from 'drizzle-orm';
-import { auth } from '$lib/auth';
+import { tryGetAuthenticatedUserId } from '$lib/backend/auth-utils';
 
 const COST_PER_MINUTE: Record<string, number> = {
 	'Whisper-Large-V3': 0.01,
 	Wizper: 0.01,
 	'Elevenlabs-STT': 0.03,
 };
 
+const getApiKey = (request: Request): string | null => {
+	const authHeader = request.headers.get('Authorization');
+	if (authHeader?.startsWith('Bearer ')) {
+		const token = authHeader.slice(7).trim();
+		if (token.length > 0) {
+			return token;
+		}
+	}
+
+	return request.headers.get('x-api-key') || env.NANOGPT_API_KEY || null;
+};
+
 export const POST: RequestHandler = async ({ request, fetch }) => {
 	try {
 		const formData = await request.formData();
@@ -31,7 +43,7 @@ export const POST: RequestHandler = async ({ request, fetch }) => {
 			formData.set('language', language);
 		}
 
-		const apiKey = request.headers.get('x-api-key') || env.NANOGPT_API_KEY;
+		const apiKey = getApiKey(request);
 
 		if (!apiKey) {
 			return json({ error: 'API key is required' }, { status: 401 });
@@ -70,9 +82,8 @@ export const POST: RequestHandler = async ({ request, fetch }) => {
 		// Track analytics asynchronously
 		(async () => {
 			try {
-				const session = await auth.api.getSession({ headers: request.headers });
-				if (session?.user?.id) {
-					const userId = session.user.id;
+				const userId = await tryGetAuthenticatedUserId(request);
+				if (userId) {
 					const durationMs = Date.now() - start;
 
 					// Try to get cost from various response headers

src/routes/api/tts/+server.ts

@@ -4,7 +4,7 @@ import type { RequestHandler } from './$types';
 import { db } from '$lib/db';
 import { modelPerformanceStats } from '$lib/db/schema';
 import { eq, and, sql } from 'drizzle-orm';
-import { auth } from '$lib/auth';
+import { tryGetAuthenticatedUserId } from '$lib/backend/auth-utils';
 
 const COST_PER_1K_CHARS: Record<string, number> = {
 	'gpt-4o-mini-tts': 0.0125,
@@ -17,6 +17,18 @@ const COST_PER_1K_CHARS: Record<string, number> = {
 // Default to standard price if unknown
 const DEFAULT_COST = 0.015;
 
+const getApiKey = (request: Request): string | null => {
+	const authHeader = request.headers.get('Authorization');
+	if (authHeader?.startsWith('Bearer ')) {
+		const token = authHeader.slice(7).trim();
+		if (token.length > 0) {
+			return token;
+		}
+	}
+
+	return request.headers.get('x-api-key') || env.NANOGPT_API_KEY || null;
+};
+
 export const POST: RequestHandler = async ({ request, fetch }) => {
 	let textStr = '';
 	let modelId = 'tts-1';
@@ -31,7 +43,7 @@ export const POST: RequestHandler = async ({ request, fetch }) => {
 			return json({ error: 'Text is required' }, { status: 400 });
 		}
 
-		const apiKey = request.headers.get('x-api-key') || env.NANOGPT_API_KEY;
+		const apiKey = getApiKey(request);
 
 		if (!apiKey) {
 			return json({ error: 'API key is required' }, { status: 401 });
@@ -61,6 +73,12 @@ export const POST: RequestHandler = async ({ request, fetch }) => {
 			);
 		}
 
+		const contentType = response.headers.get('Content-Type') || '';
+		if (response.status === 202 || contentType.includes('application/json')) {
+			const data = await response.json();
+			return json(data, { status: response.status });
+		}
+
 		// Return the audio blob directly
 		console.log(
 			`[TTS] API response status: ${response.status}, headers:`,
@@ -71,9 +89,8 @@ export const POST: RequestHandler = async ({ request, fetch }) => {
 		// Track analytics asynchronously
 		(async () => {
 			try {
-				const session = await auth.api.getSession({ headers: request.headers });
-				if (session?.user?.id) {
-					const userId = session.user.id;
+				const userId = await tryGetAuthenticatedUserId(request);
+				if (userId) {
 					console.log(`[TTS] Tracking analytics for user: ${userId}, model: ${model}`);
 					const duration = Date.now() - start;
 

src/routes/api/tts/status/+server.ts

@@ -0,0 +1,54 @@
+import { json } from '@sveltejs/kit';
+import type { RequestHandler } from './$types';
+
+const getApiKey = (request: Request): string | null => {
+	const authHeader = request.headers.get('Authorization');
+	if (authHeader?.startsWith('Bearer ')) {
+		const token = authHeader.slice(7).trim();
+		if (token.length > 0) {
+			return token;
+		}
+	}
+
+	return request.headers.get('x-api-key') || env.NANOGPT_API_KEY || null;
+};
+
+export const GET: RequestHandler = async ({ request, fetch, url }) => {
+	const apiKey = getApiKey(request);
+
+	if (!apiKey) {
+		return json({ error: 'API key is required' }, { status: 401 });
+	}
+
+	const runId = url.searchParams.get('runId');
+	const model = url.searchParams.get('model');
+
+	if (!runId || !model) {
+		return json({ error: 'runId and model are required' }, { status: 400 });
+	}
+
+	const proxyUrl = new URL('https://nano-gpt.com/api/tts/status');
+	for (const [key, value] of url.searchParams.entries()) {
+		proxyUrl.searchParams.set(key, value);
+	}
+
+	const response = await fetch(proxyUrl.toString(), {
+		headers: {
+			'x-api-key': apiKey,
+		},
+	});
+
+	const contentType = response.headers.get('Content-Type') || '';
+	if (contentType.includes('application/json')) {
+		const data = await response.json();
+		return json(data, { status: response.status });
+	}
+
+	const text = await response.text();
+	return new Response(text, {
+		status: response.status,
+		headers: {
+			'Content-Type': contentType,
+		},
+	});
+};