cmusphinx/pocketsphinx

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

4335 stars · 732 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

PocketSphinx is Carnegie Mellon's lightweight, speaker-independent continuous speech recognition engine, available as a C library, a command-line tool, and a Python module. It uses older but compact and efficient algorithms, making it well suited to resource-constrained and offline applications.

Use cases

• transcribe speech from wav files offline
• add speech recognition to a raspberry pi project
• embed a small speech recognizer in a C application
• recognize speech in python without cloud services
• run keyword spotting on an embedded device
• convert spoken audio to text locally

When to choose

• you need fully offline, low-resource speech recognition
• you're targeting embedded or low-power hardware like Raspberry Pi
• you want a simple C API or Python bindings without heavy dependencies
• accuracy requirements are modest and compactness matters more than state-of-the-art results

When to avoid

• you need state-of-the-art accuracy on noisy or diverse audio
• you want modern neural models or multilingual support out of the box
• you need streaming recognition with robust audio capture built in (the audio library was removed)
• you're building a production voice assistant where Whisper-class models are feasible

Facets

library · maturity active

speech-recognition cli nlp speech-processing embedded-systems windows python c cli embedded offline-speech-recognition asr cmu-sphinx lightweight embedded-speech natural-language-processing linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="cmusphinx/pocketsphinx")