pocketsphinx
Official Python bindings for PocketSphinx
Decision gist · record as of 2026-08-14
Yes, if you need offline speech recognition without cloud dependencies and can accept that the engine is mature but no longer state-of-the-art. Install friction is moderate (system library required on Linux/macOS), but prebuilt wheels ease setup. No security vulnerabilities are known. Best suited for keyword spotting, voice commands, and prototyping rather than high-accuracy transcription tasks.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- PortAudio library must be installed on the system (libportaudio2 on Debian-like systems) for LiveSpeech to work; AudioFile class does not require it.
- Medium install friction due to compiled wheels and a system dependency: PortAudio (libportaudio2 on Debian) is required for the LiveSpeech class to function.
- Prebuilt wheels are available for recent Python versions on common platforms.
License · maintenance · safety
permissive license (permissive) — Licensed under a permissive BSD-like license, which allows commercial and private use with minimal restrictions. No licensing concerns for typical adoption.
last release 2026-06-06 (69 days) · last repo commit 2026-08-10 · 4,332 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 382,553 downloads/mo, #7,086 on PyPI
Alternatives
Verify before relying
pip install pocketsphinx
from pocketsphinx import LiveSpeech
for phrase in LiveSpeech():
print(phrase)- Whether the package's accuracy and model quality meet requirements for production speech recognition tasks.
- Performance characteristics and latency for real-time transcription on typical hardware.
- Availability and quality of language models beyond the default US English model.
What it is and what it does
PocketSphinx is a mature, speaker-independent continuous speech recognition engine originally developed at Carnegie Mellon University. It provides Python bindings to recognize speech from live microphone streams or pre-recorded audio files, supporting both general speech-to-text transcription and keyword spotting. The package includes built-in language models and dictionaries for US English and can be extended with custom models.
The library offers two main interfaces: LiveSpeech for real-time microphone input and AudioFile for processing recorded audio. It depends on sounddevice for audio I/O and requires the PortAudio library on most systems. Development has largely ceased and the engine is acknowledged to be far from state-of-the-art, but it remains actively maintained and is used in production by many projects.
Use it for
- Build offline speech-to-text applications that don't require cloud API calls or internet connectivity.
- Implement keyword spotting or wake-word detection in IoT or embedded voice applications.
- Process batch audio files to extract transcriptions with custom language models.
- Create voice command interfaces for desktop or server applications with low latency requirements.
- Prototype speech recognition features without dependency on external speech recognition services.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you need offline speech recognition without cloud dependencies and can accept that the engine is mature but no longer state-of-the-art.
Install friction is moderate (system library required on Linux/macOS), but prebuilt wheels ease setup. No security vulnerabilities are known. Best suited for keyword spotting, voice commands, and prototyping rather than high-accuracy transcription tasks.
Install
pocketsphinx on PyPI
Before you install
Medium install friction due to compiled wheels and a system dependency: PortAudio (libportaudio2 on Debian) is required for the LiveSpeech class to function. Prebuilt wheels are available for recent Python versions on common platforms. The package is actively maintained with a recent release (69 days old) and 4332 repository stars.
PortAudio library must be installed on the system (libportaudio2 on Debian-like systems) for LiveSpeech to work; AudioFile class does not require it.
License in practice
Licensed under a permissive BSD-like license, which allows commercial and private use with minimal restrictions. No licensing concerns for typical adoption.
Quickstart
pip install pocketsphinx
from pocketsphinx import LiveSpeech
for phrase in LiveSpeech():
print(phrase)Verify before relying
- Whether the package's accuracy and model quality meet requirements for production speech recognition tasks.
- Performance characteristics and latency for real-time transcription on typical hardware.
- Availability and quality of language models beyond the default US English model.
Package facts
| License | permissive license permissive |
| Python support | Not specified |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 1 packagesounddevice |
| Maintenance | Actively maintained 69 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 382,553 / month, #7,086 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Development Status :: 6 - MatureLicense :: OSI Approved :: BSD LicenseOperating System :: OS IndependentProgramming Language :: CProgramming Language :: CythonProgramming Language :: Python :: 3Programming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Programming Language :: Python :: 3.13Programming Language :: Python :: 3.9Topic :: Multimedia :: Sound/Audio :: Speech |
Evidence: pocketsphinx-5.1.1-cp311-cp311-macosx_10_9_universal2.whl; pocketsphinx-5.1.1-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl; pocketsphinx-5.1.1-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl; pocketsphinx-5.1.1-cp311-cp311-musllinux_1_2_aarch64.whl; pocketsphinx-5.1.1-cp311-cp311-musllinux_1_2_x86_64.whl; pocketsphinx-5.1.1-cp311-cp311-win_amd64.whl; pocketsphinx-5.1.1-cp313-cp313-macosx_10_13_universal2.whl; pocketsphinx-5.1.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl; pocketsphinx-5.1.1-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl; pocketsphinx-5.1.1-cp313-cp313-musllinux_1_2_aarch64.whl; pocketsphinx-5.1.1-cp313-cp313-musllinux_1_2_x86_64.whl; pocketsphinx-5.1.1-cp313-cp313-win_amd64.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 › “speech recognition python”
- pocketsphinxPocketSphinx provides Python bindings for Carnegie Mellon…
- voskVosk provides offline speech recognition for 20+ languages, enabling…
- SpeechRecognitionPerforms speech recognition and transcription using multiple online…
Give your agent the search over MCP, or paste the wish link into any chat.
More Speech packages
Performs speech recognition and transcription using multiple online and offline engines, including Google, OpenAI Whisper, CMU Sphinx, and others.
The main gotcha is that most engines require optional dependencies or API credentials.
gTTS converts text to speech using Google Translate's API, writing MP3 audio to files, file-like objects, or stdout via Python library or command-line tool.
However, be aware that it depends on Google Translate's undocumented API—upstream changes can break it without notice, and it is not a substitute for official Google…
Lhotse prepares multimodal (speech, audio, video, image, text) data for machine learning model training with flexible pipelines, on-the-fly augmentation, and efficient data loading.
Install it if you are building speech, audio, or multimodal training pipelines; skip it if you only need simple audio I/O without data augmentation or complex dataset…
Piper TTS is a local neural text-to-speech engine that converts text to speech using embedded phonemization, with support for multiple languages and voices.
FunASR is a speech recognition toolkit that transcribes audio offline or via streaming, with integrated voice activity detection, speaker identification, punctuation restoration, and emotion/audio-event tagging across multiple languages and deployment targets.
Install it if you need speaker diarization, emotion detection, streaming support, or self-hosted deployment.
Python bindings for ai-coustics audio enhancement, voice activity detection, and analysis SDK, supporting real-time audio processing with numpy arrays.
See also sherpa-onnx · sherpa-onnx-core · onnx-asr · openwakeword · vosk · pyobjc-framework-Speech · deepgram-sdk · qwen-asr · whisperx