# microsoft/SpeechT5

Unified-Modal Speech-Text Pre-Training for Spoken Language Processing

Repository: https://github.com/microsoft/SpeechT5
Canonical: https://ross.abutalabs.com/products/speecht5
Language: Python
License: MIT
License Family: permissive
Topics: speech-pretraining, speech2c, speecht5, speechlm, speechut, speech-recognition, speech-synthesis, speech-translation, speech-text-pretraining, vallex, vatlm
Last push: 2024-04-24T05:27:36+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1667, "days_push": 861, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1449, forks 135 (observed 2026-08-28T04:04:45.691972+00:00)

## What it is
Microsoft's unified-modal speech-text pre-training framework implementing SpeechT5 and related models (SpeechLM, SpeechUT, VALL-E X, WavLLM) for speech recognition, synthesis, and translation. It provides research code, pretrained models, and HuggingFace integration for spoken language processing tasks.

## Use cases
- convert text to speech with a pretrained tts model
- transcribe speech to text with asr
- translate speech directly to another language
- clone a voice across languages
- pretrain speech and text models jointly
- run speecht5 models from huggingface

## When to choose
- you need research-grade pretrained models for ASR, TTS, or speech translation
- you want to reproduce or build on Microsoft's speech-text pre-training papers
- you want HuggingFace-compatible SpeechT5 checkpoints

## When to avoid
- you need a production-ready, well-supported TTS/ASR product
- you want simple inference without GPU resources or research setup
- you need actively maintained code with frequent updates

## Facets
- artifact type: library
- maturity: maintenance
- function: speech-recognition, tts, machine-learning, deep-learning, transformers
- domain: speech-processing, machine-learning, artificial-intelligence
- platform: python
- tags: speech-text-pretraining, speech-synthesis, speech-translation, asr, research-code, huggingface, natural-language-processing, linux, gpu

## Member repositories
- microsoft/SpeechT5 (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:45.691972+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 2026-08-30T04:36:00.420604+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/SpeechT5 (fetched 2026-08-28T04:04:45.691972+00:00, sha 87a89a946437)
- Data as of 2026-08-30T08:39:29.467469+00:00.