# NVIDIA/audio-flamingo

PyTorch implementation of Audio Flamingo: Series of Advanced Audio Understanding Language Models

Repository: https://github.com/NVIDIA/audio-flamingo
Canonical: https://ross.abutalabs.com/products/audio-flamingo
Homepage: https://research.nvidia.com/labs/adlr/AF3/
License Family: other
Topics: audio-captioning, audio-language-models, multimodal-large-language-models, audio-question-answering, audio-reasoning
Last push: 2025-12-15T18:24:26+00:00

## Health v2 (maintenance only)
Score: 50/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 57, release rhythm 35, longevity 59
- inputs: {"age_days": 835, "days_push": 261, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1182, forks 104 (observed 2026-08-28T04:03:54.254542+00:00)

## What it is
NVIDIA's PyTorch implementation of the Audio Flamingo series of large audio-language models (AF1, AF2, AF3, and Music Flamingo) for audio understanding, captioning, question answering, and reasoning. It includes model checkpoints, training recipes, and datasets for long-audio understanding, multi-turn audio chat, and voice-to-voice interaction.

## Use cases
- caption audio files automatically
- ask questions about audio recordings
- build an audio question answering model
- understand long audio up to 10 minutes
- generate music and song descriptions
- run a multimodal audio chat model locally
- fine-tune an audio language model on my own data
- reason about speech and sounds with an LLM

## When to choose
- you need state-of-the-art open audio understanding with QA, captioning, or reasoning
- you want to research or fine-tune large audio-language models with released training recipes and datasets
- you need long-audio (up to 10 minutes) comprehension or multi-turn audio chat
- you have GPU resources and want fully open checkpoints

## When to avoid
- you need a lightweight production speech-to-text service rather than a research LALM
- you lack GPU hardware, since these are multi-billion-parameter models
- you need a permissively licensed dependency - the repo lists no explicit license
- you only need simple audio feature extraction or classic ASR

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, audio-processing, speech-recognition, nlp, llm-inference, llm-training
- domain: artificial-intelligence, large-language-models, speech-processing, machine-learning
- platform: python
- tags: audio-language-model, multimodal, audio-captioning, audio-question-answering, pytorch, nvidia, music-understanding, research-models, audio, gpu, linux

## Member repositories
- NVIDIA/audio-flamingo (main) score 50

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:54.254542+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 2026-08-30T06:24:46.112424+00:00, confidence not recorded.
  - readme: https://github.com/NVIDIA/audio-flamingo (fetched 2026-08-28T04:03:54.254542+00:00, sha 309654aa17e8)
  - homepage: https://research.nvidia.com/labs/adlr/AF3/ (fetched 2026-08-29T12:32:05.313597+00:00, sha e706548cb762)
  - site_page: https://www.nvidia.com/en-us/about-nvidia/privacy-policy (fetched 2026-08-29T12:32:05.317117+00:00, sha 5362c58d0750)
  - site_page: https://www.nvidia.com/en-us/about-nvidia/privacy-center (fetched 2026-08-29T12:32:05.319873+00:00, sha b098377da9cc)
  - site_page: https://www.nvidia.com/en-us/about-nvidia/terms-of-service (fetched 2026-08-29T12:32:05.321643+00:00, sha 85469b6ff0a1)
  - site_page: https://www.nvidia.com/en-us/about-nvidia/accessibility (fetched 2026-08-29T12:32:05.324086+00:00, sha 8f61ce4143ec)
  - site_page: https://www.nvidia.com/en-us/about-nvidia/company-policies (fetched 2026-08-29T12:32:05.325997+00:00, sha b169611fd7d6)
- Data as of 2026-08-30T08:39:29.467469+00:00.