@lalalic/local-vision-audio

extension

Local vision (describe images), STT (speech-to-text), and TTS (text-to-speech). macOS uses MLX (Apple Silicon GPU). Windows/Linux use whisper, edge-tts, and transformers.

by · v1.0.0 · published 1mo ago

$ pi install npm:@lalalic/local-vision-audio
downloads/mo
0
stars
last push
open issues

Signals

license: MITtestspi manifest: missinginstall size: —deps: 0peer deps: 0

Download trend

185 downloads · last 12 weeks (weekly)

README

@lalalaic/local-vision-audio

Local vision, STT, and TTS tools for pi. Cross-platform: macOS (Apple Silicon MLX) and Windows/Linux (whisper, edge-tts, transformers).

pi install npm:@lalalaic/local-vision-audio

Tools

describe_image(path, prompt?, model?)

PlatformEngineDetails
macOSmlx-vlmGPU-accelerated, models up to 72B
Windows/Linuxtransformers + torchBLIP image captioning

transcribe_audio(path, model?, language?)

PlatformEngineDetails
macOSmlx-audio STTWhisper, VibeVoice, Qwen3-ASR
Windows/Linuxopenai-whisperStandard whisper models

text_to_speech(text, voice?, ref_audio?, instruct?, ...)

PlatformEngineDetails
macOSmlx-audio TTSVoice cloning + design, VoxCPM2/Kokoro
Windows/Linuxedge-ttsMicrosoft Edge TTS engine

Self-healing errors

When a CLI or Python module is missing, the tool prints platform-specific install commands instead of a cryptic error:

mlx_vlm.generate not found.
Install mlx-vlm:
  uv tool install mlx-vlm
  Or: pip install mlx-vlm

Requirements

macOS (Apple Silicon)

uv tool install mlx-vlm     # Vision
uv tool install mlx-audio   # STT + TTS

Windows / Linux

pip install openai-whisper          # STT
pip install edge-tts                # TTS
pip install transformers torch torchvision Pillow  # Vision (CPU/GPU)

License

MIT