tts-train
This skill provides a general, composable workflow for building datasets and training
npx skills add majiayu000/claude-skill-registry --skill tts-train --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
--- name: tts-train description: > Build TTS datasets and train Qwen3-TTS voice models from audiobooks or curated clips. Use for voice cloning, dataset prep, WhisperX alignment, Qwen3-TTS training, and inference. allowed-tools: Bash, Read triggers: - tts train - train voice model - voice cloning - voice model training - tts dataset - build tts dataset - align transcripts - qwen3 tts training - qwen tts training - tts fine tune metadata: short-description: End-to-end TTS dataset prep + training workflow for Qwen3-TTS provides: - tts-train composes: [, task-monitor] --- # TTS Train Workflow This skill provides a general, composable workflow for building datasets and training Qwen3-TTS voice models. It has its own `pyproject.toml` (self-contained) for Qwen3-TTS + TensorBoard. Use the bundled `run.sh` so the correct environment is selected per step: - **Project env** for ingest/alignment (uses `whisperx` + `faster-whisper`). - **Skill env** for Qwen3-TTS training/inference + TensorBoard (avoids pandas conflicts). ```bash .agent/skills/tts-train/run.sh <command> ... ``` ## Dataset Build Options ### Option A: Audiobook → Segments (fast path) Use the audiobook-ingest skill for ingestion,
- Dataset Build Options
- Option A: Audiobook → Segments (fast path)
- Option B: Curated Clips + Transcripts
- Alignment (WhisperX)
- Iterative Training Loop (Recommended)
- Why Two Phases?
- Full Workflow (Recommended)
- Evaluation-Only (Skip Hyperparameter Search)
- Custom Evaluation Phrases
- How It Works
- Outputs
- Monitor Hyperparameter Search
- Training (Qwen3-TTS)
- TTS Architecture
Full workflow: hyperparameter search + iterative training python .agent/skills/tts-train/iterative_train.py \ With hyperparams file Auto-evaluation mode (no manual rating) View Optuna dashboard optuna-dashboard sqlite:///artifacts/tts/<voice>_iterative/hyperparameter_search/optuna_study.db Convert existing manifest to Qwen3-TTS format with audio_codes Train Qwen3-TTS model cd third_party/Qwen3-TTS git status # Should show modified sft_12hz.py
What does the tts-train skill do?
This skill provides a general, composable workflow for building datasets and training
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill tts-train --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.