Voice and music · by Neywa Labs (formerly Blaizzy)
MLX-Audio
Clone voices, speak text and transcribe audio, all on your Mac.
speech → text, data · text, speech → speech · audio → data, series
- Way in
- Official APINo sign-up, no key
- To start
- Freeopen source (MIT); model licenses vary
- Runs
- On your Mac
- Checked
- Nov 3, 2026Prices as of that day
What it can do 4
Transcribe speech locally (Whisper, Parakeet v3, Qwen3-ASR, Voxtral Realtime, Nemotron 3.5 ASR, VibeVoice-ASR, Granite Speech)speech → text, datafree (local)
Synthesize or clone speech locally (Kokoro, Qwen3-TTS, Chatterbox v3, VoxCPM2, Voxtral TTS, CSM, Dia, Phonon, Higgs v3, ...)16 GB: stay with <= ~4B TTS models at 4-8 bittext, speech → speechfree (local)
Detect speech segments and speakers in audioaudio → data, seriesfree (local)
Run open TTS models locally on the Mac’s GPU via MLXtext, speech → speechfree
Does the same job
Prices and limits as of Nov 3, 2026, from Neywa Labs (formerly Blaizzy)’s own pages. They change; check before you rely on them.Not affiliated with Neywa Labs (formerly Blaizzy). Not advice.Something wrong?