Modern ASR

A unified, extensible, future-proof toolkit for locally running state-of-the-art LLM-based ASR models.

简体中文 · Features · Installation · Models · Quick Start · Architecture

✨ Features

🧩 19 Models — Whisper, SenseVoice, Qwen, MiMo, FireRedASR, GLM-ASR, and more.
🔌 Zero-Config Plugin — Add new models via @register_model decorator.
🚀 Runtime Hot-Swap — Switch models without restarting the process.
🌍 Multi-Language — 52 languages, 22 Chinese dialects.
🎯 Multi-Task — Transcription, translation, diarization, emotion, events.
💻 Local-First — All inference on-device. No API keys. No data leaves your machine.
🍎 Apple Silicon — Native MPS (Metal Performance Shaders) support.
📦 Auto-Install — Dependencies, git repos, and HF weights are installed automatically on first use.
🐍 Modern Python — Pydantic configs, rich CLI, ISO-timestamped logging.

📦 Installation

pip install modern-asr

Dependencies and model weights are installed automatically the first time you use a model — just type its name:

from modern_asr import ASRPipeline

pipe = ASRPipeline("sensevoice-small")
pipe = ASRPipeline("mimo-asr-v2.5")
pipe = ASRPipeline("whisper-small")

For offline/air-gapped environments, pre-install everything:

pip install modern-asr[all-models]

Available extras: transformers, vllm, onnx, firered-asr, sensevoice, fun-asr, qwen-asr, mimo-asr, glm-asr, whisper, moonshine, all-models, all-backends, all.

Requirements: Python ≥ 3.10.

🧩 Supported Models

Series	Model ID	Params	Languages	Extra
Whisper (OpenAI)	`whisper-tiny`	39M	99+	`whisper`
	`whisper-base`	74M	99+	`whisper`
	`whisper-small`	244M	99+	`whisper`
	`whisper-medium`	769M	99+	`whisper`
	`whisper-large-v3`	1.5B	99+	`whisper`
	`whisper-large-v3-turbo`	809M	99+	`whisper`
SenseVoice (Alibaba)	`sensevoice-small`	234M	zh/en/ja/ko/yue	`sensevoice`

# List all available models
python -m modern_asr list

🚀 Quick Start

from modern_asr import ASRPipeline

# Chinese with SenseVoice
pipe = ASRPipeline("sensevoice-small")
result = pipe("audio.wav", language="zh")
print(result.text)

# Switch to Qwen3-ASR for dialects
pipe.switch_model("qwen3-asr-0.6b")
result = pipe("audio.wav", language="zh")
print(result.text)

# English with Whisper
pipe.switch_model("whisper-small")
result = pipe("audio.wav", language="en")
print(result.text)

🏗️ Architecture

Modern ASR is built on three layers:

ASRPipeline — Unified user API. Input normalization, task dispatch, model lifecycle.
ASRModel / AudioLLMModel — Adapter layer. New models often need only 8 lines of config.
Backends — Transformers, vLLM, ONNX Runtime.

Adding a New Model

from modern_asr.core.audio_llm import AudioLLMModel
from modern_asr.core.registry import register_model

@register_model("my-model-1b")
class MyModel1B(AudioLLMModel):
    HF_PATH = "org/MyModel-1B"
    SUPPORTED_LANGUAGES = {"zh", "en"}
    CHUNK_DURATION = 30.0

    @property
    def model_id(self) -> str:
        return "my-model-1b"

The registry auto-discovers it at runtime. That's it.

📚 Documentation

Full documentation with Material for MkDocs:

mkdocs serve

🤝 Contributing

See Contributing Guide for development setup, code style, and PR checklist.

📄 License

Apache-2.0

Metadata

Release files for modern-asr 0.2.18

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for modern-asr 0.2.18
File	Size	Uploaded	Provenance
modern_asr-0.2.18.tar.gz	390.3 kB	May 3, 2026	Without provenance	Details

Built distribution (wheel)

Table of built distributions (wheels) for modern-asr 0.2.18
File	Interpreter	ABI	Platform	Reset
modern_asr-0.2.18-py3-none-any.whl 63.8 kB May 3, 2026 Without provenance	Python 3	none	any	Details

Total release size: 454.1 kB

Release files / modern_asr-0.2.18.tar.gz

Download URL	modern_asr-0.2.18.tar.gz
Size	390.3 kB
Tags	Source
SHA-256 checksum How to use checksums	`1a504ee7604002667862994f7be623f438165ee1959a98ae653d78ccc9a8c23a`
BLAKE2b-256 checksum How to use checksums	`a52ebdda67d8d18a6f61fa8123cb7a183b5d43456931263b71b62e95dbda0693`
Upload date	May 3, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`uv/0.11.8 {"installer":{"name":"uv","version":"0.11.8","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}`

Release files / modern_asr-0.2.18-py3-none-any.whl

Download URL	modern_asr-0.2.18-py3-none-any.whl
Size	63.8 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`3d9ec9cdacbf12eceac2fffb53dcd3dedee719651c49e09d054500dc2dd35fc2`
BLAKE2b-256 checksum How to use checksums	`5aca35f5e23db93f3d1637551a82bf12a3a87035492d2e99bf404359ab1b8ddb`
Upload date	May 3, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`uv/0.11.8 {"installer":{"name":"uv","version":"0.11.8","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}`

Security for modern-asr 0.2.18

Report project as malware

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.