kaldi-asr/kaldi

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

15469 stars · 5354 forks observed · 2026-08-28

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

Kaldi is a C++ toolkit for speech recognition research and development, including acoustic modeling, feature extraction, decoding, and speaker identification. It ships with example recipe scripts and supports GPU (CUDA) acceleration for training neural network models.

Use cases

• build a speech-to-text system
• train acoustic models for ASR
• speaker identification and verification
• transcribe audio recordings offline
• research HMM and DNN speech recognition
• keyword spotting in audio

When to choose

• you need a mature, research-grade ASR toolkit with full training pipelines
• you want fine-grained control over acoustic modeling and decoding
• you need speaker recognition in addition to transcription
• you can work with C++ and shell scripts on Linux/Unix

When to avoid

• you want a simple pretrained model API with minimal setup (consider Vosk or Whisper)
• you need Python-first tooling or quick prototyping
• you are building mobile or browser-based real-time transcription
• you cannot maintain shell-script-based recipe workflows

Facets

library · maturity maintenance

speech-recognition machine-learning deep-learning audio-processing gpu-computing speech-processing machine-learning windows cpp asr speech-to-text speaker-id kaldi research-toolkit c-plus-plus natural-language-processing audio linux macos cuda gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="kaldi-asr/kaldi")