Convert audio for speech recognition
Whisper, DeepSpeech and most ASR pipelines expect 16 kHz mono PCM and reject or mis-transcribe anything else — resampled-by-hand files still fail on channel count or bit depth. One pass delivers the exact spec: 16000 Hz, one channel, signed 16-bit little-endian WAV.
Drop your file here — or click to choose
audio file (MP3/WAV/M4A/FLAC/OGG/AAC) · up to 200MB each · processing starts instantly, preview is free
No signup needed. Only pay if you want the clean file.
Frequently asked questions
Why 16 kHz and not 44.1 kHz?
Speech models are trained on telephone-band audio (16 kHz); higher rates just waste 3× the disk and RAM without improving transcripts.
Does it work with stereo files?
Yes — channels are mixed down to mono automatically.
When this tool is not the right choice
You need a different sample rate for delivery
use the sample rate converter
use the sample rate converter