# janhq/ichigo

Local realtime voice AI

Repository: https://github.com/janhq/ichigo
Canonical: https://ross.abutalabs.com/products/ichigo
Language: Python
License Family: other
Last push: 2025-11-26T22:03:38+00:00

## Health v2 (maintenance only)
Score: 50/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 54, release rhythm 40, longevity 56
- inputs: {"age_days": 792, "days_push": 280, "days_rel": 611, "gap_med": 27.0, "n_releases_24m": 3}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2492, forks 149 (observed 2026-08-28T04:06:56.492484+00:00)

## What it is
Ichigo is a Python speech package for developers offering local realtime voice AI capabilities, including a compact 22M-parameter speech tokenizer (Ichigo-ASR) built on Whisper-medium and an experimental speech language model (Ichigo-LLM). It provides inference-only code with intuitive Python interfaces and a scalable FastAPI service for speech tasks.

## Use cases
- transcribe audio to text locally
- run speech recognition offline
- build a realtime voice assistant
- convert speech into discrete tokens for LLMs
- batch process audio files for transcription
- serve speech recognition via a FastAPI endpoint
- experiment with speech language models

## When to choose
- you need local, privacy-preserving speech recognition without cloud APIs
- you want a lightweight ASR model that integrates with LLMs via discrete tokens
- you need multilingual transcription (English, Vietnamese) with a small model
- you want both a Python API and a FastAPI service for speech tasks

## When to avoid
- you need production-grade text-to-speech today (TTS is listed as coming soon)
- you require a commercially licensed dependency (no license is specified)
- you need highly accurate ASR across many languages beyond English and Vietnamese
- you want training code rather than inference only

## Facets
- artifact type: library
- maturity: active
- function: speech-recognition, tts, llm-inference, audio-processing, machine-learning
- domain: speech-processing, artificial-intelligence
- platform: python, cross-platform
- tags: asr, voice-ai, local-inference, speech-tokens, fastapi, whisper, natural-language-processing, audio, gpu

## Member repositories
- janhq/ichigo (main) score 50

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:56.492484+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:27:45.878909+00:00, confidence not recorded.
  - readme: https://github.com/janhq/ichigo (fetched 2026-08-28T04:06:56.492484+00:00, sha 5df21b0ee64c)
- Data as of 2026-08-30T08:39:29.467469+00:00.