silero-vad
Voice Activity Detector (VAD) by Silero
Decision gist · record as of 2026-08-14
Yes. Active maintenance, no known vulnerabilities, permissive MIT license, low install friction, and production-grade accuracy make it a solid choice for any speech detection task. The main gotcha is ensuring an audio backend (FFmpeg, sox, or soundfile) is available on your deployment target—verify that before committing to it in a containerized or embedded environment.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires torch>=1.12.0 and torchaudio>=0.12.0; also needs an audio backend (FFmpeg, sox, or soundfile) installed on the system for audio I/O.
- Low friction: pure Python wheel with no compiled dependencies beyond torch and torchaudio.
- Active maintenance with recent commits and strong repository engagement.
License · maintenance · safety
permissive license (permissive) — MIT license (permissive) means you can use, modify, and distribute freely with minimal restrictions. No telemetry, vendor lock-in, or expiration built in.
last release 2026-02-24 (171 days) · last repo commit 2026-07-16 · 9,945 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 1,260,439 downloads/mo, #4,150 on PyPI
Alternatives
Verify before relying
pip install silero-vad
from silero_vad import load_silero_vad, read_audio, get_speech_timestamps
model = load_silero_vad()
wav = read_audio('audio.wav')
speech_timestamps = get_speech_timestamps(wav, model, return_seconds=True)- Whether the package's claimed support for over 6000 languages is validated independently or represents training data scope only.
- Exact CPU instruction set requirements (AVX, AVX2, AVX-512, AMX) and fallback behavior on older hardware.
- Whether audio backend (FFmpeg, sox, or soundfile) installation is automatic or manual for each deployment target.
What it is and what it does
Silero VAD is a pre-trained neural network model that identifies when speech is present in audio. It takes audio input and returns timestamps marking where voice activity occurs, useful for filtering silence, segmenting conversations, or triggering downstream processing only when speech is detected.
The package wraps PyTorch models (or ONNX alternatives) and handles audio loading via torchaudio, supporting 8000 Hz and 16000 Hz sampling rates. It's designed for production use: the model is around two megabytes, processes audio chunks in under 1ms on CPU, and was trained on diverse multilingual data. You can run it on CPU, GPU, or via ONNX runtime on various architectures.
Use it for
- Filter silence from voice recordings or call center logs before transcription or analysis.
- Trigger voice bot responses only when speech is detected, reducing false activations.
- Segment long audio files into speech regions for downstream processing (transcription, speaker diarization).
- Implement voice interfaces on edge devices or mobile by detecting when a user starts speaking.
- Clean training datasets by identifying and extracting only speech-containing portions.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes.
Active maintenance, no known vulnerabilities, permissive MIT license, low install friction, and production-grade accuracy make it a solid choice for any speech detection task. The main gotcha is ensuring an audio backend (FFmpeg, sox, or soundfile) is available on your deployment target—verify that before committing to it in a containerized or embedded environment.
Install
silero-vad on PyPI
Before you install
Low friction: pure Python wheel with no compiled dependencies beyond torch and torchaudio. Active maintenance with recent commits and strong repository engagement.
Requires torch>=1.12.0 and torchaudio>=0.12.0; also needs an audio backend (FFmpeg, sox, or soundfile) installed on the system for audio I/O.
License in practice
MIT license (permissive) means you can use, modify, and distribute freely with minimal restrictions. No telemetry, vendor lock-in, or expiration built in.
Quickstart
pip install silero-vad
from silero_vad import load_silero_vad, read_audio, get_speech_timestamps
model = load_silero_vad()
wav = read_audio('audio.wav')
speech_timestamps = get_speech_timestamps(wav, model, return_seconds=True)Verify before relying
- Whether the package's claimed support for over 6000 languages is validated independently or represents training data scope only.
- Exact CPU instruction set requirements (AVX, AVX2, AVX-512, AMX) and fallback behavior on older hardware.
- Whether audio backend (FFmpeg, sox, or soundfile) installation is automatic or manual for each deployment target.
Package facts
| License | permissive license permissive |
| Python support | Supports the current Python release >=3.8 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 3 packagespackagingtorchtorchaudio |
| Maintenance | Actively maintained 171 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 1,260,439 / month, #4,150 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Development Status :: 5 - Production/StableIntended Audience :: DevelopersIntended Audience :: Science/ResearchLicense :: OSI Approved :: MIT LicenseOperating System :: OS IndependentProgramming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Programming Language :: Python :: 3.13Programming Language :: Python :: 3.14Programming Language :: Python :: 3.15Programming Language :: Python :: 3.8Programming Language :: Python :: 3.9Topic :: Scientific/EngineeringTopic :: Scientific/Engineering :: Artificial Intelligence |
Evidence: silero_vad-6.2.1-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “VAD pre-trained model”
- silero-vadSilero VAD detects speech activity in audio files and streams,…
- openwakewordopenWakeWord detects wake words and phrases in audio streams using…
- realtimesttRealtimeSTT converts speech from microphone or external audio streams…
Give your agent the search over MCP, or paste the wish link into any chat.
More Scientific/Engineering packages
NumPy provides an N-dimensional array object and a comprehensive suite of mathematical, linear algebra, Fourier transform, and random number functions for scientific computing in Python.
pandas provides fast, flexible data structures (Series and DataFrame) for loading, cleaning, transforming, and analyzing labeled or relational data in Python.
scipy provides numerical algorithms for mathematics, science, and engineering—including optimization, integration, linear algebra, Fourier transforms, signal and image processing, and ODE solvers—built on numpy arrays.
scikit-learn provides a comprehensive Python library for supervised and unsupervised machine learning, including classification, regression, clustering, dimensionality reduction, and model evaluation tools built on NumPy and SciPy.
Install it if you need to train, evaluate, or deploy supervised or unsupervised learning models.
dill extends Python's pickle module to serialize and deserialize a much wider range of Python objects, including functions, lambdas, classes, and interpreter sessions, to byte streams for storage or network transmission.
Multiprocess is an enhanced fork of Python's standard multiprocessing library that uses dill for better serialization, allowing you to spawn processes with a threading-like API and share complex objects between them.
Install it if you use multiprocessing and encounter pickle serialization limits with lambdas or complex objects.
See also silero · webrtcvad · webrtcvad-wheels · pymicro-vad · sherpa-onnx · openwakeword · realtimestt · pyloudnorm · onnx-asr · sherpa-onnx-core