Skip to content

Voice-to-text dictation app with local (Nvidia Parakeet/Whisper) and cloud models (BYOK). Privacy-first and available cross-platform.

fix: always pre-convert audio to 16kHz mono WAV for whisper-server

Gabriel Stein5daf5627f1e4
diff --git a/src/helpers/whisperServer.js b/src/helpers/whisperServer.js
index b0c4da179..18e7a5ec7 100644
--- a/src/helpers/whisperServer.js
+++ b/src/helpers/whisperServer.js
@@ -6,7 +6,7 @@ const http = require("http");
const debugLogger = require("./debugLogger");
const { killProcess } = require("../utils/process");
const { getSafeTempDir } = require("./safeTempDir");
-const { isWavFormat, convertToWav } = require("./ffmpegUtils");
+const { convertToWav } = require("./ffmpegUtils");
const PORT_RANGE_START = 8178;
const PORT_RANGE_END = 8199;
@@ -223,14 +223,13 @@ class WhisperServerManager {
const args = ["--model", modelPath, "--host", "127.0.0.1", "--port", String(this.port)];
- // Only add --convert flag if FFmpeg is available
+ // FFmpeg is required for pre-converting audio to 16kHz mono WAV
this.canConvert = !!ffmpegPath;
if (ffmpegPath) {
- args.push("--convert");
const ffmpegDir = path.dirname(ffmpegPath);
spawnEnv.PATH = ffmpegDir + pathSep + spawnEnv.PATH;
} else {
- debugLogger.warn("FFmpeg not found - whisper-server will only accept WAV format");
+ debugLogger.warn("FFmpeg not found - whisper-server will only accept 16kHz mono WAV");
}
if (options.threads) args.push("--threads", String(options.threads));
@@ -385,15 +384,12 @@ class WhisperServerManager {
const { language, initialPrompt } = options;
- // Pre-convert to WAV so whisper-server doesn't create temp files internally
+ // Always convert to 16kHz mono WAV - whisper.cpp requires this exact format
let finalBuffer = audioBuffer;
- const isWav = isWavFormat(audioBuffer);
- if (!isWav) {
- if (!this.canConvert) {
- throw new Error("FFmpeg not found - whisper-server requires WAV input");
- }
- finalBuffer = await this._convertToWav(audioBuffer);
+ if (!this.canConvert) {
+ throw new Error("FFmpeg not found - required for audio conversion");
}
+ finalBuffer = await this._convertToWav(audioBuffer);
const boundary = `----WhisperBoundary${Date.now()}`;
const parts = [];