VoiceKit MCP

by brickhousestudiosllcco.huggingface.brickhousestudiosllc/voicekit

voicekit_analyze_acoustics Analyze audio with normalized PCM + real DSP metrics (real-dsp-v1).
voicekit_analyze_acoustics_halves Normalize input once, split normalized PCM 50/50, then run real DSP on each half.
voicekit_analyze_notes Estimate note-like pitched segments from normalized PCM using real DSP (real-dsp-v1).
voicekit_extract_embedding Extract a deterministic local voice embedding (demo mode).
voicekit_grade_voice Compare two voices/audio clips using real DSP feature similarity (real-dsp-v1). This is not ASR pronunciation grading yet; pronunciation/transcript fields are retained for M4/UI compatibility and marked as acoustic proxies.
voicekit_isolate_voice Separate vocals/instrumental with Demucs and persist artifacts to the mounted HF bucket.
voicekit_match_voice Compare two audios via deterministic local embeddings (demo mode).
voicekit_read_artifact_chunk Read a bounded private Demucs stem chunk when the exact source hash is known.
voicekit_transcribe_audio Placeholder transcription lane. Real ASR is not wired in real-dsp-v1.

As the server's own listing gives them. We never call a tool.

Tools