{"data":{"slug":"espnet-espnet","name":"espnet","tagline":"End-to-End Speech Processing Toolkit","github_url":"https://github.com/espnet/espnet","owner":"espnet","repo":"espnet","owner_avatar_url":"https://avatars.githubusercontent.com/u/34493687?v=4","primary_language":"Python","stars":9963,"forks":2435,"topics":["chainer","deep-learning","end-to-end","kaldi","machine-translation","pytorch","singing-voice-synthesis","speaker-diarization","speech-enhancement","speech-recognition","speech-separation","speech-synthesis","speech-translation","spoken-language-understanding","text-to-speech","voice-conversion"],"archived":false,"github_pushed_at":"2026-09-20T05:11:59+00:00","maintenance_label":"Very active","stars_delta_30d":60,"url":"https://www.graphcanon.com/tools/espnet-espnet","markdown_url":"https://www.graphcanon.com/tools/espnet-espnet.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/espnet-espnet","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=espnet-espnet","description":"End-to-End Speech Processing Toolkit","homepage_url":"https://espnet.github.io/espnet/","license":"Apache-2.0","open_issues":79,"watchers":171,"ai_summary":"ESPNet offers tools for speech recognition, synthesis, translation, and more tasks employing deep learning models via Python.","readme_excerpt":"# Install PyTorch first: https://pytorch.org/get-started/locally/\npip install espnet              # run pretrained models\npip install \"espnet[train]\"     # also train them, with espnet2 or espnet3 (Lightning, TensorBoard, W&B, Hydra, Dask, ...)\n```\n\n<details>\n<summary>Other installation options</summary>\n\n```sh\npip install \"espnet[all]\"                       # training plus every task extra (except sds)\npip install git+https://github.com/espnet/espnet  # latest master\n```\n\n- **Full setup** (recipes, DNN training, Kaldi-style tooling): see the\n  [installation guide](https://espnet.github.io/espnet/installation.html).\n- **Docker**: see [`docker/`](docker/) and the [Docker docs](https://espnet.github.io/espnet/docker.html).\n- **Task-specific tools** live in [`tools/installers`](tools/installers).\n- **ESPnet1 is no longer supported** — use ESPnet2 (`egs2/`) or ESPnet3 (`egs3/`). See [the ESPnet1 notice](https://espnet.github.io/espnet/espnet1_tutorial.html).\n\n</details>\n\n<details>\n<summary>Tested environments (CI status)</summary>\n\n|system/pytorch ver.|2.11.0|2.13.0|2.14.0|\n| :---- | :---: | :---: | :---: |\n|ubuntu/python3.12/pip||||\n|ubuntu/python3.13/pip||||\n|debian12/python3.12/conda||||\n|windows/python3.12/pip||||\n|macos/python3.12/pip||||\n|macos/python3.12/conda||||\n\nEach badge is its workflow's aggregate status on `master`, where the full grid runs.\nCoverage is not uniform — some suites run on one pytorch only, and a pull request runs\nless than `master` does. [What each column covers](CONTRIBUTING.md#53-what-runs-on-a-pull-request).\n\n\n\n\n\n\n</details>\n\n---\n\n## Quick start\n\n**From the terminal** — no code, on any audio file:\n\n```sh\npip install espnet\nespnet asr audio.wav                       # transcribe, detecting the language\nespnet translate audio.wav --to eng        # speech in, English text out\nespnet tts \"Hello from ESPnet\" -o out.wav\nespnet enhance noisy.wav -o clean.wav\nespnet asr --live                          # transcribe the microphone\nespnet models                              # the default model of each command\npip install \"espnet[demo]\"\nespnet demo                                # the same model in your browser\n```\n\nEvery command takes `--model <tag>` and `--device cuda`, and `espnet --version` names\nthe installed version. `espnet demo` serves, on localhost, the app behind [the OWSM-CTC\nv4 Space](https://huggingface.co/spaces/espnet/owsm-ctc-v4) — recording, upload, the\ncheckpoint's own language and translation menus, and long-form decoding — and takes\n`--port` and `--share`.\n\n**Without installing anything** — the same commands, in a container:\n\n```sh\ndocker run --rm -v \"$PWD:/data\" -v \"$HOME/.cache/huggingface:/cache/huggingface\" \\\n    espnet/espnet:inference-cpu-latest asr /data/audio.wav\n```\n\nWith a GPU, `espnet/espnet:inference-gpu-latest`, `--gpus all` and\n`--device cuda`.\n\nThe second mount is what keeps the downloaded model between runs; on a Linux\nhost add `--user \"$(id -u):$(id -g)\"`, so that what it writes belongs to you.\nThe other two images, and what each is for, are in [`docker/`](docker/).\n\n**From Python** — any model from the [ESPnet Hugging Face organization](https://huggingface.co/espnet):\n\n```python\nfrom espnet2.bin.s2t_inference import Speech2Text","github_created_at":"2017-12-13T00:45:11+00:00","created_at":"2026-07-11T12:05:26.447004+00:00","updated_at":"2026-09-20T07:04:29.301762+00:00","categories":[{"slug":"model-training","name":"Model Training","url":"https://www.graphcanon.com/categories/model-training","markdown_url":"https://www.graphcanon.com/categories/model-training.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/model-training"},{"slug":"speech-audio","name":"Speech & Audio","url":"https://www.graphcanon.com/categories/speech-audio","markdown_url":"https://www.graphcanon.com/categories/speech-audio.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/speech-audio"}],"tags":[{"slug":"chainer","name":"chainer"},{"slug":"deep-learning","name":"deep-learning"},{"slug":"kaldi","name":"kaldi"},{"slug":"pytorch","name":"pytorch"},{"slug":"speech-recognition","name":"speech recognition"},{"slug":"speech-synthesis","name":"speech-synthesis"},{"slug":"speech-translation","name":"speech-translation"},{"slug":"spoken-language-understanding","name":"spoken-language-understanding"}],"trust":{"provenance":{"is_fork":false,"github_id":114054873,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-09-20T07:04:27.060Z","maintenance":{"label":"Very active","score":96,"methodology":"github_public_v1","releases_90d":2,"days_since_push":0,"last_release_at":"2026-09-19T03:11:29Z","stars_delta_30d":60,"open_issues_delta_30d":30},"security_summary":{"status":"no_lockfile","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-07-11T12:05:28.195Z","medium_count":0,"scan_profile":"none","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-09-20T07:04:28.119Z"},"has_cli":{"value":true,"source":"pyproject.toml:[project.scripts]","observed_at":"2026-09-20T07:04:28.119Z"},"languages":{"value":["python"],"source":"github.language+pyproject.toml","observed_at":"2026-09-20T07:04:28.119Z"},"license_spdx":{"value":"Apache-2.0","source":"github.license","observed_at":"2026-09-20T07:04:28.119Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":null,"constraints":null,"when_to_use":["When you require comprehensive tools for end-to-end speech processing tasks such as speech recognition, synthesis, translation, and speaker diarization.","If your project involves working with Chainer or PyTorch frameworks for experimenting with DNN training in the domain of speech technology.","You need to use W&B (Weights & Biases) for tracking runs and prefer a toolkit that supports integration directly through configuration."],"when_not_to_use":["If you are working on tasks unrelated to speech or audio processing, such as computer vision, NLP, or any other deep learning areas outside of ESPNet's focus.","Your development environment is limited to languages other than Python or frameworks that do not support Chainer or PyTorch, which are foundational to espnet."],"source":"enrich:decision_facts","observed_at":"2026-07-17T13:29:08.555Z"},"constraint_facets":null,"decision_summary":[{"label":"Adopt for","value":"ESPNet is an End-to-End Speech Processing Toolkit that employs deep learning models for tasks including speech recognition and synthesis."}]}}