# wwbin2017/bailing

百聆 是一个类似GPT-4o的语音对话机器人，通过ASR+LLM+TTS实现，集成DeepSeek R1等优秀大模型，接入openClaw，真正的个人语音助手，时延低至800ms，Mac等低配置也可运行，支持打断

Repository: https://github.com/wwbin2017/bailing
Canonical: https://ross.abutalabs.com/products/bailing
Language: Python
License: MIT
License Family: permissive
Topics: ai, asr, chatgpt, chattts, deepseek, funasr, gpt-4o, llm, openai, tts, voice-assistant, openclaw
Last push: 2026-04-06T04:21:13+00:00

## Health v2 (maintenance only)
Score: 54/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 76, release rhythm 28, longevity 52
- inputs: {"age_days": 738, "days_push": 149, "days_rel": 460, "gap_med": 77, "n_releases_24m": 2}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1757, forks 306 (observed 2026-08-28T04:05:32.320435+00:00)

## What it is
Bailing is an open-source voice assistant application similar to GPT-4o, built with an ASR + VAD + LLM + TTS pipeline (FunASR, silero-vad, DeepSeek, edge-tts/Kokoro/ChatTTS). It achieves ~800ms end-to-end latency, runs without a GPU on low-end hardware like Macs, and supports interruptions, memory, tool calling via OpenClaw, and task management.

## Use cases
- build a personal voice assistant like GPT-4o
- run a low-latency voice chatbot on a Mac without GPU
- create a speech-to-speech AI assistant with interruption support
- add voice interaction with tool calling to an LLM
- self-host an open-source voice assistant with DeepSeek
- build a JARVIS-style assistant that can execute tasks
- experiment with ASR + LLM + TTS pipelines

## When to choose
- you want a self-hosted, GPU-free voice assistant with sub-second latency
- you need a modular ASR/VAD/LLM/TTS stack you can swap components in
- you want voice-driven tool calling and task management via OpenClaw
- you're running on low-resource hardware like a Mac

## When to avoid
- you need a production-grade commercial voice assistant with SLAs
- you want a fully on-device offline solution without API-based LLM/TTS
- you need multilingual support beyond what the bundled models provide
- you prefer a managed cloud voice assistant rather than assembling Python components

## Facets
- artifact type: application
- maturity: active
- function: speech-recognition, tts, llm-inference, chatbot, agent-framework, audio-processing
- domain: artificial-intelligence, large-language-models, speech-processing, chatbots
- platform: python, cross-platform, self-hosted
- tags: voice-assistant, asr, vad, gpt-4o-like, low-latency, interruptible, openclaw, deepseek, funasr, chattts, ai-agents, macos

## Member repositories
- wwbin2017/bailing (main) score 54

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:32.320435+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 2026-08-30T03:27:40.895352+00:00, confidence not recorded.
  - readme: https://github.com/wwbin2017/bailing (fetched 2026-08-28T04:05:32.320435+00:00, sha d03502e694cd)
- Data as of 2026-08-30T08:39:29.467469+00:00.