# Flashlight

A C++ standalone library for machine learning

Repository: https://github.com/flashlight/flashlight
Canonical: https://ross.abutalabs.com/products/flashlight
Homepage: https://fl.readthedocs.io/en/latest/
Language: C++
License: MIT
License Family: permissive
Topics: flashlight, machine-learning, autograd, cpp, deep-learning, neural-network, ml
Last push: 2026-06-22T00:07:38+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 88, release rhythm 8, longevity 100
- inputs: {"age_days": 2822, "days_push": 73, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5466, forks 502 (observed 2026-08-28T04:09:19.564067+00:00)

## What it is
wav2letter++ is Facebook AI Research's end-to-end automatic speech recognition (ASR) toolkit written in C++. It has been consolidated into the Flashlight library's ASR application, with this repository serving as an archive of recipes and pre-trained models for reproducing ASR research papers.

## Use cases
- train an end-to-end speech recognition model
- transcribe audio to text with a pretrained ASR model
- reproduce research papers on speech recognition
- build a streaming ConvNet speech recognizer
- run lexicon-free speech recognition

## When to choose
- you need a fast C++ end-to-end ASR toolkit with pretrained models
- you want to reproduce published wav2letter/Flashlight ASR research results
- you are training ConvNet-based speech recognition models on GPU

## When to avoid
- you need a Python-first ASR framework with broad ecosystem support
- you want actively developed new features, since development moved to Flashlight
- you need exact reproducibility without pinning Flashlight <= 0.3.2

## Facets
- artifact type: library
- maturity: maintenance
- function: speech-recognition, machine-learning, deep-learning, audio-processing
- domain: speech-processing, machine-learning, deep-learning
- platform: cpp
- tags: asr, wav2letter, end-to-end, convnets, pretrained-models, facebook-ai-research, linux, macos, gpu

## Member repositories
- flashlight/flashlight (main) score 62
- flashlight/wav2letter (mirror) score 64

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:19.564067+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:44:43.127309+00:00, confidence not recorded.
  - readme: https://github.com/flashlight/flashlight (fetched 2026-08-28T04:09:19.564067+00:00, sha a69c9cd5116a)
- Data as of 2026-08-30T08:39:29.467469+00:00.