Ffvoice

Install

Works with Claude Code, Codex, VS Code, Cursor, Windsurf, Claude Desktop. Read the repository before you give any tool access to your data.

npx agentmarket install connector/chicogong-ffvoice

About

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list and install config. Checked Oct 25, 2026.

Tools

3 read · 0 write · 0 delete · 2 not yet classified

Secrets

None asked for

Install scripts

Not scanned yet

Runs as

Python package

Tools (5)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• capture_and_caption

  Capture live microphone audio and produce real-time captions using LiveCaptioner. Everything runs fully on-device via Whisper ASR and VAD segmentation. No audio or transcription data is ever transmitted to any external service. The captioner emits two kinds of events while audio is being captured: - **Partial**: intermediate caption emitted at *partial_interval_ms* intervals while speech is ongoing. The text may change as more audio arrives. - **Final**: definitive caption emitted at the end of each detected utterance. Includes a non-zero confidence score. Args: duration_seconds: …

• capture_and_transcribe

  Capture live microphone audio for a given duration and transcribe it. Everything runs locally on this device — no audio is transmitted anywhere. VAD (Voice Activity Detection) segments the stream into speech chunks which are each transcribed individually by Whisper ASR. Args: duration_seconds: How many seconds to record (must be between 1 and 120). device_id: PortAudio device ID to use. Pass -1 (default) for the system default input device. denoise: Apply RNNoise AI denoising before VAD/ASR when True. If RNNoise is not available in this build the tool automatical…

• list_audio_devices

  List all audio input/output devices available on this machine. Uses PortAudio under the hood; no data leaves the device. Returns:: { "devices": [ { "id": int, "name": str, "max_input_channels": int, "max_output_channels": int, "supported_sample_rates": [int, ...], "is_default": bool }, ... ], "default_input_device_id": int }

• transcribe_file

  Transcribe a local audio file to text using offline Whisper ASR. Processing is fully on-device — no audio is sent anywhere. Supports WAV, FLAC, and any format that whisper.cpp can read. Args: path: Absolute or relative path to the audio file. language: BCP-47 language code ('en', 'zh', 'fr', …) or 'auto' for automatic language detection. model: Whisper model size — one of 'tiny', 'base', 'small', 'medium', 'large'. Larger models are more accurate but slower. word_timestamps: When True, each segment includes per-word timing in the 'words' field. Retur…

• transcribe_file_with_diarization

  Transcribe a local audio file and label each segment with a speaker. Combines offline Whisper ASR with speaker diarization ("who spoke when"): the file is transcribed, diarized, and every transcription segment is annotated with the speaker who was talking during it. Processing is fully on-device — no audio is sent anywhere. Diarization works whenever either the compiled C++ diarizer is present (source build with ENABLE_DIARIZATION=ON) or the ``sherpa-onnx`` PyPI package is installed (``pip install 'ffvoice[diarization]'``). When neither is available it returns a structured error dict inste…

Signals

Stars

3

Forks

0

Open issues

0

Weekly downloads

66

Last commit

Aug 26, 2026 (2 months ago)

Last release

May 19, 2026 (5 months ago)

Last package publish

May 19, 2026 (5 months ago)

From GitHub (fetched Oct 24, 2026) and the package registry.

Details

Registry name

io.github.chicogong/ffvoice

Version

0.8.3

Auth

No auth

Packages

pypi ffvoice

License

MIT

Repository

github.com/chicogong/ffvoice-engine

First seen

Oct 24, 2026

Last checked

Oct 25, 2026

Source

Official MCP Registrygithub-graphql