# HumeAI/tada

Open Source Speech Language Model

Repository: https://github.com/HumeAI/tada
Canonical: https://ross.abutalabs.com/products/tada
Homepage: https://www.hume.ai/blog/opensource-tada
Language: Jupyter Notebook
License: NOASSERTION
License Family: other
Last push: 2026-05-11T17:59:55+00:00

## Health v2 (maintenance only)
Score: 51/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 81, release rhythm 35, longevity 12
- inputs: {"age_days": 179, "days_push": 114, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1009, forks 112 (observed 2026-08-28T04:03:12.846260+00:00)

## What it is
TADA is an open-source speech-language model from Hume AI that generates expressive, high-fidelity speech via text-acoustic dual alignment, synchronizing text and audio tokens one-to-one. It is released as a Python package (hume-tada) with pre-trained models on Hugging Face and claims to be the fastest LLM-based TTS system with virtually zero content hallucination.

## Use cases
- convert text to natural expressive speech
- generate voice audio from a script
- build a voice agent with fast TTS
- run text-to-speech on-device with low memory
- synthesize emotional or expressive speech
- avoid hallucinated or skipped words in TTS output

## When to choose
- you need fast, reliable LLM-based text-to-speech with low hallucination
- you want expressive, natural-sounding speech generation in Python
- you need efficient inference with bf16, prompt caching, and torch.compile optimizations
- you want an open-source TTS model light enough for on-device deployment

## When to avoid
- you need speech recognition or transcription rather than speech synthesis
- you need a fully managed TTS API rather than self-hosted model inference
- you require a permissively licensed model without restrictions (license is non-standard)
- you need non-English speech synthesis support

## Facets
- artifact type: library
- maturity: active
- function: tts, speech-recognition, llm-inference, machine-learning, deep-learning
- domain: speech-processing, large-language-models, artificial-intelligence
- platform: python, cross-platform
- tags: speech-language-model, text-to-speech, voice-generation, generative-ai, hugging-face, on-device-ai, natural-language-processing, gpu

## Member repositories
- HumeAI/tada (main) score 51

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:12.846260+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 2026-08-30T07:12:00.525645+00:00, confidence not recorded.
  - readme: https://github.com/HumeAI/tada (fetched 2026-08-28T04:03:12.846260+00:00, sha 2ac78574aba9)
  - homepage: https://www.hume.ai/blog/opensource-tada (fetched 2026-08-29T13:12:12.693877+00:00, sha 1e3d144399d6)
  - site_page: https://www.hume.ai/about (fetched 2026-08-29T13:12:12.703705+00:00, sha 74a0c114192f)
  - site_page: https://dev.hume.ai (fetched 2026-08-29T13:12:12.705779+00:00, sha 9d8aa4d851e1)
- Data as of 2026-08-30T08:39:29.467469+00:00.