yangtan501-collab

Vocal Coach — AI skill for Claude Code

AI community

🎤 EVT + CVT 双体系 AI 声乐教练 — Claude Code agent skill.

How to install Vocal Coach

This entry records only its repository, not the path inside it, so there is no exact command to give. Open yangtan501-collab/vocal-coach and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Vocal Coach does

🎤 EVT + CVT 双体系 AI 声乐教练 — Claude Code agent skill.

Alternatives in AI

Prompt Coach Skill — Interactive prompt coaching and improvement 135 ★
2026 Coach — Executive coaching skill for planning your 2026 using research-backed process goals 33 ★
Critical Thinking For Humans — A gym for your critical thinking — the AI is the coach, you do the work 32 ★

README

🎤 Vocal Coach / 声乐教练

**EVT + CVT dual-system AI vocal coach for Claude Code.**

Analyzes spectrograms, diagnoses voice issues using Estill Voice Training (EVT) and Complete Vocal Technique (CVT), then gives exercises you can actually *feel* in your body.

"Don't compare against perfection — compare against *this person's own baseline*. Anatomical differences are not defects."

[中文说明](#中文说明) ↓

Quickstart (2 minutes)

# 1. Install in Claude Code
npx skills add yangtan501-collab/vocal-coach@vocal-coach

# 2. Install Python deps
pip install -r ~/.claude/skills/vocal-coach/requirements.txt

# 3. Optional: visual analysis engine (pick one)
#    Mac Apple Silicon:
pip install mlx mlx-vlm
#    Windows / Linux / any GPU:
#    Install Ollama → ollama pull qwen3-vl

Then in Claude Code:

/vocal-coach
/vocal-coach analyze this-recording.wav

What It Does

Dual-Engine Spectrogram Analysis

Every recording is analyzed by two independent systems, cross-validated:

Engine	Tech	Role
Engine A — Numbers	Python (Parselmouth/Praat)	Extract pitch, throat space, tongue position, striation, brightness, per-segment stats
Engine B — Vision	MLX / Ollama + Vision LLM	Directly looks at the spectrogram image, describes what it sees

Two engines, blind to each other. Agreement = high confidence. Disagreement = needs more investigation.

Full CVT + EVT Dimension Sweep

CVT 4 Modes — Neutral / Curbing / Overdrive / Edge detection
CVT Metal Density — from breathy to piercing
EVT 13 Figures — every structure checked, no dimension skipped:
- Fig 1: Vocal fold onset · Fig 2: False vocal folds · Fig 3: Vocal fold thickness
- Fig 4: Larynx position · Fig 5: Soft palate · Fig 6: Larynx tilt
- Fig 7: Tongue · Fig 8: Jaw · Fig 9: Lips
- Fig 10: AES (singer's formant) · Fig 12: Head/neck anchoring · Fig 13: Torso anchoring