Add audio conversion endpoint and client integration

Author: richiemcilroy

apps/media-server/src/app.ts

@@ -21,6 +21,7 @@ app.get("/", (c) => {
 			"/audio/status",
 			"/audio/check",
 			"/audio/extract",
+			"/audio/convert",
 			"/video/status",
 			"/video/probe",
 			"/video/thumbnail",

apps/media-server/src/routes/audio.ts

@@ -19,6 +19,12 @@ const extractSchema = z.object({
 	stream: z.boolean().optional().default(true),
 });
 
+const convertSchema = z.object({
+	audioUrl: z.string().url(),
+	outputFormat: z.literal("mp3").optional().default("mp3"),
+	bitrate: z.string().optional().default("128k"),
+});
+
 function isBusyError(err: unknown): boolean {
 	return err instanceof Error && err.message.includes("Server is busy");
 }
@@ -173,4 +179,73 @@ audio.post("/extract", async (c) => {
 	}
 });
 
+audio.post("/convert", async (c) => {
+	const body = await c.req.json();
+	const result = convertSchema.safeParse(body);
+
+	if (!result.success) {
+		return c.json(
+			{
+				error: "Invalid request",
+				code: "INVALID_REQUEST",
+				details: result.error.message,
+			},
+			400,
+		);
+	}
+
+	const { audioUrl, bitrate } = result.data;
+
+	try {
+		const { stream, cleanup } = extractAudioStream(audioUrl, {
+			bitrate,
+			timeoutMs: 15 * 60 * 1000,
+		});
+
+		c.req.raw.signal.addEventListener("abort", () => {
+			cleanup();
+		});
+
+		return new Response(stream, {
+			headers: {
+				"Content-Type": "audio/mpeg",
+				"Transfer-Encoding": "chunked",
+			},
+		});
+	} catch (err) {
+		console.error("[audio/convert] Error:", err);
+
+		if (isBusyError(err)) {
+			return c.json(
+				{
+					error: "Server is busy",
+					code: "SERVER_BUSY",
+					details: "Too many concurrent requests, please retry later",
+				},
+				503,
+			);
+		}
+
+		if (isTimeoutError(err)) {
+			return c.json(
+				{
+					error: "Request timed out",
+					code: "TIMEOUT",
+					details: err instanceof Error ? err.message : String(err),
+				},
+				504,
+			);
+		}
+
+		return c.json(
+			{
+				error: "Failed to convert audio",
+				code: "FFMPEG_ERROR",
+				details: err instanceof Error ? err.message : String(err),
+			},
+			500,
+		);
+	}
+});
+
 export default audio;

apps/web/lib/audio-enhance.ts

@@ -1,7 +1,9 @@
-import { spawn } from "node:child_process";
 import { serverEnv } from "@cap/env";
 import Replicate from "replicate";
-import { getFfmpegPath } from "./audio-extract";
+import {
+	convertAudioToMp3ViaMediaServer,
+	isMediaServerConfigured,
+} from "./media-client";
 
 const MAX_POLL_ATTEMPTS = 120;
 const POLL_INTERVAL_MS = 5000;
@@ -10,51 +12,7 @@ export const ENHANCED_AUDIO_EXTENSION = "mp3";
 export const ENHANCED_AUDIO_CONTENT_TYPE = "audio/mpeg";
 
 export function isAudioEnhancementConfigured(): boolean {
-	return !!serverEnv().REPLICATE_API_TOKEN;
-}
-
-async function streamConvertToMp3(wavUrl: string): Promise<Buffer> {
-	const ffmpeg = getFfmpegPath();
-	const ffmpegArgs = [
-		"-i",
-		wavUrl,
-		"-acodec",
-		"libmp3lame",
-		"-b:a",
-		"128k",
-		"-f",
-		"mp3",
-		"-pipe:1",
-	];
-
-	return new Promise((resolve, reject) => {
-		const proc = spawn(ffmpeg, ffmpegArgs, { stdio: ["pipe", "pipe", "pipe"] });
-
-		const chunks: Buffer[] = [];
-		let stderr = "";
-
-		proc.stdout?.on("data", (chunk: Buffer) => {
-			chunks.push(chunk);
-		});
-
-		proc.stderr?.on("data", (data: Buffer) => {
-			stderr += data.toString();
-		});
-
-		proc.on("error", (err: Error) => {
-			reject(new Error(`Audio conversion failed: ${err.message}`));
-		});
-
-		proc.on("close", (code: number | null) => {
-			if (code === 0) {
-				resolve(Buffer.concat(chunks));
-			} else {
-				reject(
-					new Error(`Audio conversion failed with code ${code}: ${stderr}`),
-				);
-			}
-		});
-	});
+	return !!serverEnv().REPLICATE_API_TOKEN && isMediaServerConfigured();
 }
 
 export async function enhanceAudioFromUrl(audioUrl: string): Promise<Buffer> {
@@ -63,6 +21,10 @@ export async function enhanceAudioFromUrl(audioUrl: string): Promise<Buffer> {
 		throw new Error("REPLICATE_API_TOKEN is not configured");
 	}
 
+	if (!isMediaServerConfigured()) {
+		throw new Error("MEDIA_SERVER_URL is not configured");
+	}
+
 	const replicate = new Replicate({
 		auth: apiToken,
 	});
@@ -106,7 +68,7 @@ export async function enhanceAudioFromUrl(audioUrl: string): Promise<Buffer> {
 		throw new Error("No output received from Replicate");
 	}
 
-	const mp3Buffer = await streamConvertToMp3(enhancedAudioUrl);
+	const mp3Buffer = await convertAudioToMp3ViaMediaServer(enhancedAudioUrl);
 
 	return mp3Buffer;
 }

apps/web/lib/media-client.ts

@@ -142,3 +142,39 @@ export async function extractAudioViaMediaServer(
 	const arrayBuffer = await response.arrayBuffer();
 	return Buffer.from(arrayBuffer);
 }
+
+export async function convertAudioToMp3ViaMediaServer(
+	audioUrl: string,
+): Promise<Buffer> {
+	const mediaServerUrl = serverEnv().MEDIA_SERVER_URL;
+	if (!mediaServerUrl) {
+		throw new Error("MEDIA_SERVER_URL is not configured");
+	}
+
+	const response = await fetchWithRetry(`${mediaServerUrl}/audio/convert`, {
+		method: "POST",
+		headers: { "Content-Type": "application/json" },
+		body: JSON.stringify({
+			audioUrl,
+			outputFormat: "mp3",
+			bitrate: "128k",
+		}),
+	});
+
+	if (!response.ok) {
+		let errorData: MediaServerError;
+		try {
+			errorData = (await response.json()) as MediaServerError;
+		} catch {
+			throw new Error(
+				`Audio conversion failed: ${response.status} ${response.statusText}`,
+			);
+		}
+		throw new Error(
+			errorData.details || errorData.error || "Audio conversion failed",
+		);
+	}
+
+	const arrayBuffer = await response.arrayBuffer();
+	return Buffer.from(arrayBuffer);
+}