Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
{
"schema_version": 1,
"lead": "funclip-v2-2-0-moss-speaker-clipping",
"selected": [
"meeting-transcript-acceptance",
"fun-asr-nano-transformers",
"self-hosted-openai-whisper-api-alternative",
"funasr-transcribe-long-audio"
],
"lead_image": "/img/funclip-v2-1-0-interface.jpg",
"articles": [
{
"slug": "websocket-offline-model-options",
"category": "explanations",
"reviewed": true,
"zh": {
"title": "换了模型却没生效?核对 WebSocket 启动配置",
"summary": "区分离线和流式模型,用不加载权重的源码检查确认参数路由,再规划真实录音验收。"
},
"en": {
"title": "Changed the model, but not the result? Check the WebSocket configuration",
"summary": "Separate offline and streaming models, inspect argument routing without loading weights, then plan acceptance on real recordings."
}
},
{
"slug": "hotwords-package-boundaries",
"category": "explanations",
"reviewed": true,
"zh": {
"title": "热词没有生效?先核对传递链路和安装包",
"summary": "从 WebSocket 消息到 Torch 独立运行时,核对热词格式、安装版本和批次时间戳,避免把接口修复当作准确率保证。"
},
"en": {
"title": "Hotwords not working? Check the request path and installed package",
"summary": "Check hotword encoding, standalone package versions and batched timestamps, without confusing interface fixes with recognition accuracy guarantees."
}
},
{
"slug": "sensevoice-finetuning-acceptance",
"category": "explanations",
"reviewed": true,
"zh": {
"title": "微调后,怎样确认模型真的更好?",
"summary": "从一次持续学习反馈出发,先验证评测尺子,再选择 checkpoint,最后检查真实服务。"
},
"en": {
"title": "Did fine-tuning actually make your model better?",
"summary": "Validate the ruler, choose a checkpoint under old-domain constraints, then test the real service."
}
},
{
"slug": "cantonese-speech-recognition",
"category": "applications",
"reviewed": false,
"zh": {
"title": "粤语录音转写实践",
"summary": ""
},
"en": {
"title": "Transcribing Cantonese recordings",
"summary": ""
}
},
{
"slug": "chinese-speech-recognition",
"category": "applications",
"reviewed": false,
"zh": {
"title": "中文录音转写入门",
"summary": ""
},
"en": {
"title": "Getting started with Mandarin transcription",
"summary": ""
}
},
{
"slug": "fun-asr-nano-guide",
"category": "selection",
"reviewed": false,
"zh": {
"title": "Fun-ASR-Nano 使用笔记",
"summary": ""
},
"en": {
"title": "Using Fun-ASR-Nano",
"summary": ""
}
},
{
"slug": "fun-asr-nano-transformers",
"category": "selection",
"reviewed": true,
"zh": {
"title": "接入 Transformers,先选对权重",
"summary": "同名模型的权重格式不能混用。先理清格式,再跑通原生接入。"
},
"en": {
"title": "Choose the right checkpoint for Transformers",
"summary": "Checkpoint formats are not interchangeable. Choose the right artifact before connecting the API."
}
},
{
"slug": "funasr-cli-transcribe-command-line",
"category": "applications",
"reviewed": false,
"zh": {
"title": "从命令行转写音频",
"summary": ""
},
"en": {
"title": "Transcribing audio from the command line",
"summary": ""
}
},
{
"slug": "funasr-llama-cpp-whisper-cpp-alternative",
"category": "selection",
"reviewed": false,
"zh": {
"title": "用 llama.cpp 部署中文语音识别",
"summary": ""
},
"en": {
"title": "Chinese speech recognition with llama.cpp",
"summary": ""
}
},
{
"slug": "funasr-realtime-streaming-asr",
"category": "applications",
"reviewed": false,
"zh": {
"title": "实时字幕的流式识别路径",
"summary": ""
},
"en": {
"title": "A streaming path for live captions",
"summary": ""
}
},
{
"slug": "funasr-speaker-diarization",
"category": "explanations",
"reviewed": false,
"zh": {
"title": "说话人分离:谁在何时说话",
"summary": ""
},
"en": {
"title": "Speaker diarization: who spoke when",
"summary": ""
}
},
{
"slug": "funasr-transcribe-long-audio",
"category": "explanations",
"reviewed": true,
"zh": {
"title": "长录音为什么会漏掉结尾?",
"summary": "从切分边界、生成上限和结果覆盖出发,检查长录音的遗漏。"
},
"en": {
"title": "Why can a long transcript miss the ending?",
"summary": "Check segmentation, generation limits and output coverage when a recording seems incomplete."
}
},
{
"slug": "funasr-v1-3-26-openai-vllm-llama-cpp",
"category": "releases",
"reviewed": false,
"zh": {
"title": "v1.3.26:API 与运行时入口",
"summary": ""
},
"en": {
"title": "v1.3.26: API and runtime entry points",
"summary": ""
}
},
{
"slug": "funasr-v1-3-27-language-metadata-vllm-fallback",
"category": "releases",
"reviewed": false,
"zh": {
"title": "v1.3.27:语种信息与回退处理",
"summary": ""
},
"en": {
"title": "v1.3.27: Language metadata and fallback",
"summary": ""
}
},
{
"slug": "funasr-v1-3-28-realtime-websocket-subtitles",
"category": "releases",
"reviewed": false,
"zh": {
"title": "v1.3.28:实时识别与字幕修复",
"summary": ""
},
"en": {
"title": "v1.3.28: Realtime and subtitle fixes",
"summary": ""
}
},
{
"slug": "funasr-v1-4-0-pypi-release",
"category": "releases",
"reviewed": false,
"zh": {
"title": "v1.4.0:安装包与参数校验",
"summary": ""
},
"en": {
"title": "v1.4.0: Packaging and argument validation",
"summary": ""
}
},
{
"slug": "funasr-v1-4-14-portable-source-release",
"category": "releases",
"reviewed": false,
"zh": {
"title": "v1.4.14:源码包与 MOSS 入口",
"summary": ""
},
"en": {
"title": "v1.4.14: Source packages and MOSS discovery",
"summary": ""
}
},
{
"slug": "funasr-v1-4-16-submodel-device-release",
"category": "releases",
"reviewed": true,
"zh": {
"title": "v1.4.16:混合设备子模型修复",
"summary": "保持 VAD、标点和说话人子模型的显式设备配置,并补齐原生 Transformers 指南。"
},
"en": {
"title": "v1.4.16: Mixed-device submodel fix",
"summary": "Preserve explicit VAD, punctuation, and speaker placement, with tested native Transformers guides."
}
},
{
"slug": "funasr-v1-4-3-pypi-release",
"category": "releases",
"reviewed": false,
"zh": {
"title": "v1.4.3:VAD 与说话人聚类",
"summary": ""
},
"en": {
"title": "v1.4.3: VAD and speaker clustering",
"summary": ""
}
},
{
"slug": "funasr-v1-4-5-pypi-llama-cpp-release",
"category": "releases",
"reviewed": false,
"zh": {
"title": "v1.4.5:Python 依赖与运行时",
"summary": ""
},
"en": {
"title": "v1.4.5: Python dependencies and runtimes",
"summary": ""
}
},
{
"slug": "funasr-vs-faster-whisper-chinese",
"category": "selection",
"reviewed": false,
"zh": {
"title": "中文与粤语:历史对比实验",
"summary": ""
},
"en": {
"title": "Chinese and Cantonese: an earlier comparison",
"summary": ""
}
},
{
"slug": "funasr-vs-whisper-benchmark",
"category": "selection",
"reviewed": false,
"zh": {
"title": "FunASR 与 Whisper:历史测量记录",
"summary": ""
},
"en": {
"title": "FunASR and Whisper: historical measurements",
"summary": ""
}
},
{
"slug": "funclip-v2-1-0-video-clipping-release",
"category": "releases",
"reviewed": false,
"zh": {
"title": "FunClip v2.1.0:首个版本化发布",
"summary": ""
},
"en": {
"title": "FunClip v2.1.0: The first versioned release",
"summary": ""
}
},
{
"slug": "funclip-v2-2-0-moss-speaker-clipping",
"category": "applications",
"reviewed": true,
"zh": {
"title": "把多人录音变成可剪辑的字幕",
"summary": "从一段对话出发,生成带匿名说话人标签的字幕,再按人选段。"
},
"en": {
"title": "Turn a conversation into editable subtitles",
"summary": "Create subtitles with recording-local speaker labels, then select the passages you need."
}
},
{
"slug": "generate-subtitles-srt-vtt-from-audio-video",
"category": "applications",
"reviewed": true,
"zh": {
"title": "字幕文件生成了,然后呢?",
"summary": "从一段普通话录音得到 SRT,转换为 VTT,再回听检查时间与文字。"
},
"en": {
"title": "You have a subtitle file. Is it ready?",
"summary": "Create SRT from a Mandarin recording, convert it to VTT, then check timing and text against the audio."
}
},
{
"slug": "japanese-speech-recognition",
"category": "applications",
"reviewed": false,
"zh": {
"title": "日语录音转写实践",
"summary": ""
},
"en": {
"title": "Transcribing Japanese recordings",
"summary": ""
}
},
{
"slug": "lightweight-speech-recognition-cpu",
"category": "selection",
"reviewed": false,
"zh": {
"title": "CPU 语音识别的部署取舍",
"summary": ""
},
"en": {
"title": "Deployment choices for CPU speech recognition",
"summary": ""
}
},
{
"slug": "meeting-transcript-acceptance",
"category": "applications",
"reviewed": true,
"zh": {
"title": "会议转写,怎样才算做好了?",
"summary": "有文字还不够:检查关键内容、说话人轮次和时间覆盖。"
},
"en": {
"title": "When is a meeting transcript ready to use?",
"summary": "Text alone is not enough. Review key details, speaker turns and time coverage."
}
},
{
"slug": "punctuation-restoration-python",
"category": "explanations",
"reviewed": true,
"zh": {
"title": "文字都在,标点为什么还是断错?",
"summary": "保留原始转写,生成标点候选,再检查归句和专名,不把句号当作语义修复。"
},
"en": {
"title": "The words are right. Why is the punctuation wrong?",
"summary": "Keep the original, generate a candidate, then review sentence meaning and names."
}
},
{
"slug": "self-hosted-deepgram-assemblyai-alternative",
"category": "selection",
"reviewed": false,
"zh": {
"title": "迁移语音 API 前,先核对什么",
"summary": ""
},
"en": {
"title": "What to check before migrating a speech API",
"summary": ""
}
},
{
"slug": "self-hosted-google-aws-azure-speech-to-text",
"category": "selection",
"reviewed": false,
"zh": {
"title": "自托管语音服务的迁移笔记",
"summary": ""
},
"en": {
"title": "Notes on migrating to self-hosted speech",
"summary": ""
}
},
{
"slug": "self-hosted-openai-whisper-api-alternative",
"category": "applications",
"reviewed": true,
"zh": {
"title": "把语音转写接进自己的 API",
"summary": "先跑通本地请求,再明确鉴权、返回格式和服务责任。"
},
"en": {
"title": "Add transcription to your own API",
"summary": "Start with a local request, then define authentication, output formats and operating responsibilities."
}
},
{
"slug": "sensevoice-deployment-guide",
"category": "applications",
"reviewed": false,
"zh": {
"title": "SenseVoice 部署入门",
"summary": ""
},
"en": {
"title": "Getting started with SenseVoice",
"summary": ""
}
},
{
"slug": "sensevoice-emotion-language-detection",
"category": "explanations",
"reviewed": true,
"zh": {
"title": "SenseVoice 的情绪标签,该怎么用?",
"summary": "保留原始预测,区分显示文本与评测标签,不把分类结果当作人的真实情绪。"
},
"en": {
"title": "How should you use SenseVoice emotion tags?",
"summary": "Keep raw predictions, separate display from evaluation, and do not mistake a tag for a person's true state."
}
},
{
"slug": "speech-to-text-python-transcribe-audio",
"category": "applications",
"reviewed": false,
"zh": {
"title": "用 Python 转写一份录音",
"summary": ""
},
"en": {
"title": "Transcribe a recording with Python",
"summary": ""
}
},
{
"slug": "speech-to-text-timestamps-python",
"category": "applications",
"reviewed": false,
"zh": {
"title": "用时间戳定位原始录音",
"summary": ""
},
"en": {
"title": "Finding the original audio with timestamps",
"summary": ""
}
},
{
"slug": "subtitle-edit-fun-asr-sensevoice-local-subtitles",
"category": "applications",
"reviewed": false,
"zh": {
"title": "在 Subtitle Edit 中生成本地字幕",
"summary": ""
},
"en": {
"title": "Generating local subtitles in Subtitle Edit",
"summary": ""
}
},
{
"slug": "voice-activity-detection-python",
"category": "explanations",
"reviewed": false,
"zh": {
"title": "VAD 如何找到语音区间",
"summary": ""
},
"en": {
"title": "How VAD identifies speech regions",
"summary": ""
}
},
{
"slug": "which-funasr-model",
"category": "selection",
"reviewed": false,
"zh": {
"title": "FunASR 模型选型笔记",
"summary": ""
},
"en": {
"title": "Notes on choosing a FunASR model",
"summary": ""
}
}
]
}