MCP 服务器

kesha-voice-kit

Local voice toolkit over MCP: transcribe audio to text in 25 languages, synthesize speech in 9, and list available voices and languages. Runs fully on-device — no API keys, no cloud.

README

<h1 align="center">Kesha Voice Kit</h1>

Give your local tools and LLM agents a voice. Fast speech-to-text, text-to-speech, voice-activity detection, and language detection in one local-first CLI: Apple Silicon CoreML first, ONNX fallback on supported Linux/Windows builds.

Transcribe locally — 25 languages, up to ~19x faster than Whisper on Apple Silicon, ~2.5x on CPU
Speak back — text-to-speech in 9 languages
Plug into agents — ship voice workflows as CLI commands, an MCP server, an <a href="docs/openclaw.md">OpenClaw</a> skill, or a <a href="docs/hermes.md">Hermes</a> agent
Small Rust engine — single ~60MB binary, no ffmpeg, no Python, no native Node addons

Quick Start

Runtime: Bun >= 1.3.0 · Platforms: macOS arm64, Linux x64, Windows x64. Linux and Windows run the ONNX engine — everything except microphone capture (kesha record), macOS system voices, speaker diarization, and text language detection, which need Apple frameworks.

# 1. Install Bun (skip if you have it) — Linux & macOS:
curl -fsSL https://bun.sh/install | bash        # or: brew install oven-sh/bun/bun
# Windows: powershell -c "irm bun.sh/install.ps1 | iex"
# if `bun --version` fails, reload PATH: exec $SHELL -l

# 2. Install Kesha:
bun add -g @drakulavich/kesha-voice-kit
kesha --version                                 # confirms `kesha` resolved on PATH
kesha install --plan                            # preview exact download/disk sizes first — downloads nothing
kesha install        # ~2.5 GB on Linux/Windows; ~0.6 GB on Apple Silicon, whose CoreML
                      # engine uses a different, smaller model set. Explicit — never automatic.
                      # No progress bar during the model step; can take several minutes.
                      # Prefer a guided wizard? `kesha init` walks through the same choices interactively.

# 3. Transcribe:
kesha audio.ogg      # transcript to stdout

Prefer Homebrew, .deb/.rpm, Docker, or Nix? See Other install methods. Air-gapped or behind a corporate mirror? See docs/model-mirror.md.

Speech-to-text

kesha audio.ogg                            # transcribe (plain text)
kesha --format transcript audio.ogg        # text + language/confidence
kesha --format json audio.ogg              # full JSON with lang fields
kesha --json --timestamps audio.ogg        # JSON with timestamped segments
kesha --toon audio.ogg                     # compact LLM-friendly TOON
kesha status                               # show installed backend info
kesha status --disk                        # + recursive cache disk usage
kesha status --json                        # machine-readable, for scripts

Multiple files get head-style headers; stdout is the transcript, stderr is errors — pipe-friendly:

$ kesha freedom.ogg tahiti.ogg
=== freedom.ogg ===
Свободу попугаям! Свободу!

=== tahiti.ogg ===
Таити, Таити! Не были мы ни в какой Таити! Нас и тут неплохо кормят.

Record from the mic (macOS): kesha record --out hello.wav writes microphone audio to a WAV file (kesha hello.wav transcribes it). macOS prompts for microphone access on first use — grant it under System Settings → Privacy & Security → Microphone if it was denied. On Linux/Windows or headless boxes, pass any existing audio file straight to kesha instead.
Long / silence-heavy audio: install VAD (kesha install --vad); Kesha auto-uses it past 120 s. Without VAD, long audio falls back to fixed ASR chunks. See docs/vad.md.
Speaker diarization (darwin-arm64): kesha install --diarize, then kesha --json --vad --speakers meeting.m4a stamps each segment with a speaker id. Linux/Windows return a clear "darwin-arm64 only" error (#199).

Text-to-speech

Kesha speaks back in 9 languages, auto-picking the voice from the text's language. Override with --lang <code> or --voice <id>.

kesha install --tts                              # English voices; sizes differ per platform — preview: kesha install --plan
kesha install --tts en ru                        # + Russian (+~890 MB, Vosk)
kesha say "Hello, world" > hello.wav
kesha say "Привет, мир" > privet.wav             # auto-routes by language
kesha say --voice ru-vosk-m02 "Голос в текст." > ru.wav

Output formats (--format, or inferred from the --out extension):

kesha say "Hello" --out hi.wav                    # WAV (default, uncompressed)
kesha say "Hello" --format ogg-opus --out hi.ogg  # OGG/Opus — messenger voice notes
kesha say "Hello" --format flac --out hi.flac     # FLAC — lossless, plays in every browser incl. Safari/iOS

kesha say --list-voices lists what's installed. Voices, the full catalogue, macOS system voices, SSML, speaking rate (--rate, <prosody>), Russian word stress, and Russian/English abbreviation handling are all in docs/tts.md.

Languages

Speech-to-text spans 25 languages and text-to-speech covers English, Russian, and select multilingual voices — full tables with codes and flags in docs/languages.md. Audio language detection identifies 107 languages.

Performance

Up to ~19x faster than Whisper on Apple Silicon (M2), ~2.5x faster on CPU

Compared against Whisper large-v3-turbo, all engines auto-detecting language:

Benchmark: openai-whisper vs faster-whisper vs Kesha Voice Kit

Full per-file breakdown (Russian + English): BENCHMARK.md.

Other install methods

All of these install the Bun CLI wrapper; engine + models still download explicitly via kesha install.

Homebrew — brew install drakulavich/tap/kesha-voice-kit · docs/homebrew.md
Linux packages (.deb/.rpm, x64) — docs/linux-packages.md
Docker (GHCR image) — docs/docker.md
Nix (aarch64-darwin / x86_64-linux) — nix run github:drakulavich/kesha-voice-kit -- install · docs/nix-install.md
Shell completions + manpage — kesha completions bash|zsh|fish and kesha manpage print the packaged files to install wherever your shell expects them.

Integrations

MCP server — kesha mcp exposes transcribe/synthesize/list tools to any MCP client (Claude, Cursor, Codex, Gemini). Setup: docs/mcp.md.
OpenClaw — give your LLM agent ears. Install & config: docs/openclaw.md.
Hermes Agent — local STT/TTS through Hermes command providers. Setup: docs/hermes.md.
Raycast (macOS) — offline microphone dictation from the launcher: Dictate to Clipboard records with a live signal meter, auto-stops on silence, transcribes locally, and copies the text. Install from the Raycast Store · source: raycast/.
Programmatic API — @drakulavich/kesha-voice-kit/core for use inside a Bun program. See docs/api.md.

Architecture — runtime data flow, the models that ship, the CLI ↔ Rust engine boundary, model pinning, and where tests live.
Use cases — copy-paste recipes (transcribe a meeting, speak from OpenClaw, run offline, move the cache).
Product positioning — supported workflows, non-goals, maturity labels, platform matrix.
Diagnostics: kesha doctor, kesha support-bundle (redacted .tar.gz for issues), and kesha logs produce local, content-free diagnostics — see docs/diagnostic-logs.md. Every failure prints a stable error [CODE]: … line and a documented process exit code.
Scripting & CI: --json (or --toon) for machine-readable output, --quiet/-q to silence progress, and --no-color (or NO_COLOR=1) for plain logs. Colors switch off automatically when CI=true.
Privacy / Local Stats: Stats are off by default and fully local. Opt in with kesha stats enable to record content-free operational metrics in a local SQLite database — never networked, never storing audio, transcripts, text, or paths. Full commands & lifecycle: docs/local-stats.md.

Contributing

See CONTRIBUTING.md, the Roadmap (Now / Next / Later), and the Decision log (why platform/model choices were made — and reversed). Dev setup: make dev-setup (Bun, Rust, nextest, platform libs).

License

Made with 💛🩵 and 🥤 energy under MIT License