# innnky/emotional-vits

无需情感标注的情感可控语音合成模型，基于VITS

Repository: https://github.com/innnky/emotional-vits
Canonical: https://ross.abutalabs.com/products/emotional-vits
Language: Jupyter Notebook
License: MIT
License Family: permissive
Last push: 2023-03-30T11:10:17+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1493, "days_push": 1252, "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 1392, forks 170 (observed 2026-08-28T04:04:36.095963+00:00)

## What it is
Emotional VITS is a voice synthesis (TTS) model based on VITS that enables emotion-controllable speech generation without requiring manual emotion labels in the training dataset. It extracts emotion embeddings from audio using a pretrained emotion extraction model and feeds them into a modified TextEncoder, allowing inference with a reference audio to control the emotional tone of synthesized speech.

## Use cases
- synthesize speech with controllable emotion from a reference audio clip
- train an emotional TTS model on a dataset without emotion annotations
- cluster audio files by emotional similarity to pick reference clips
- fine-tune an existing VITS checkpoint to add emotion control
- build multi-speaker TTS with per-speaker emotional expression

## When to avoid
- you need text-driven emotion control using words like 'excited' or 'calm' without providing a reference audio
- you need a production-ready TTS service with an API out of the box
- you cannot provide a reference audio clip at inference time
- you need real-time low-latency synthesis on constrained hardware

## Facets
- artifact type: library
- maturity: maintenance
- function: tts, speech-recognition, machine-learning, deep-learning, audio-processing
- domain: speech-processing, machine-learning
- platform: python, cross-platform
- tags: vits, emotional-tts, voice-synthesis, emotion-embedding, speech-synthesis, jupyter-notebook, audio, natural-language-processing

## Member repositories
- innnky/emotional-vits (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:36.095963+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 2026-08-30T04:39:32.701870+00:00, confidence not recorded.
  - readme: https://github.com/innnky/emotional-vits (fetched 2026-08-28T04:04:36.095963+00:00, sha 0f322958fb10)
- Data as of 2026-08-30T08:39:29.467469+00:00.