Text-to-speech uses Kokoro-82M through kokoro-mlx. It runs in process after install and does not require a local TTS server. Turn it on from the menu bar or Settings -> Voice. The first enable downloads the Kokoro voice model and uses the spaCy en_core_web_sm dictionary plus system espeak-ng.

​
Use it

SurfaceCommand
KeyboardSelect text and press Option+T. Press Option+T again, Esc, or start recording to stop.
CLIwh whisper "text"
CLI with voicewh whisper --voice af_bella "text"
stdinecho "hello" | wh whisper
The overlay shows generation and speaking states.

​
Voices

VoiceIDType
Heartaf_heartAmerican female
Bellaaf_bellaAmerican female
Novaaf_novaAmerican female
Skyaf_skyAmerican female, default
Sarahaf_sarahAmerican female
Nicoleaf_nicoleAmerican female
Alicebf_aliceBritish female
Emmabf_emmaBritish female
Adamam_adamAmerican male
Echoam_echoAmerican male
Ericam_ericAmerican male
Liamam_liamAmerican male
Danielbm_danielBritish male
Georgebm_georgeBritish male

​
Configuration

[tts]
enabled = false
provider = "kokoro"
speak_shortcut = "alt+t"

[kokoro_tts]
model = "mlx-community/Kokoro-82M-bf16"
voice = "af_sky"
The provider registry lives in src/whisper_voice/tts/__init__.py.