Synthesize Speech

Generate speech from text using the Kokoro TTS CLI tool. Supports single WAV output or chunked streaming for long text.

Self-Evolving Skill: This skill improves through use. If instructions are wrong, parameters drifted, or a workaround was needed — fix this file immediately, don't defer. Only update for real, reproducible issues.

Quick Usage

# Single WAV
~/.local/share/kokoro/.venv/bin/python ~/.local/share/kokoro/tts_generate.py \
  --text "Hello from Kokoro TTS" --voice af_heart --lang en-us --speed 1.0 \
  --output /tmp/kokoro-tts-$$.wav

# Play it
afplay /tmp/kokoro-tts-$$.wav

Parameters

Parameter	Default	Description
--text	(required)	Text to synthesize
--voice	af_heart	Voice name (see voice catalog)
--lang	en-us	Language code (en-us, zh, ja, etc.)
--speed	1.0	Speech speed multiplier
--output	(required)	Output WAV path
--chunk	off	Chunked streaming mode for long text

Voice Catalog

See Voice Catalog for all available voices with quality grades.

Top voices:

Voice ID	Name	Grade	Gender
af_heart	Heart	A	Female
af_bella	Bella	A-	Female
af_nicole	Nicole	B-	Female

Chunked Streaming

For long text, use --chunk to get progressive playback:

~/.local/share/kokoro/.venv/bin/python ~/.local/share/kokoro/tts_generate.py \
  --text "Long text here..." --voice af_heart --lang en-us --speed 1.0 \
  --output /tmp/kokoro-tts-$$.wav --chunk

Each chunk WAV path is printed to stdout as it becomes ready. The final line is DONE <ms>.

Troubleshooting

Issue	Cause	Solution
No audio output	Model not loaded	Run /kokoro-tts:install first
Empty text error	Input was blank	Provide non-empty --text
Slow generation	First-run warmup	Normal — subsequent runs faster

Post-Execution Reflection

After this skill completes, check before closing:

1. Did the command succeed? — If not, fix the instruction or error table that caused the failure.
2. Did parameters or output change? — If the underlying tool's interface drifted, update Usage examples and Parameters table to match.
3. Was a workaround needed? — If you had to improvise (different flags, extra steps), update this SKILL.md so the next invocation doesn't need the same workaround.

Only update if the issue is real and reproducible — not speculative.