# memoavatar/memo

[TMLR] Memory-Guided Diffusion for Expressive Talking Video Generation

Repository: https://github.com/memoavatar/memo
Canonical: https://ross.abutalabs.com/products/memoavatar-memo
Homepage: https://memoavatar.github.io
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2025-08-06T15:27:37+00:00

## Health v2 (maintenance only)
Score: 40/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 35, release rhythm 35, longevity 62
- inputs: {"age_days": 869, "days_push": 392, "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 1070, forks 105 (observed 2026-08-28T04:03:27.897377+00:00)

## What it is
MEMO is an open-weight diffusion model for generating expressive, identity-consistent talking videos from a single reference image and an audio track. It uses memory-guided temporal modeling and emotion-aware audio attention, with inference and finetuning scripts provided in Python.

## Use cases
- generate a talking video from a photo and audio clip
- animate a portrait to match speech audio
- create expressive talking avatars with lip sync
- make a digital human sing along to a song
- finetune a talking-head model on my own video dataset
- research audio-driven video diffusion models

## When to choose
- you need state-of-the-art expressive talking video generation from a single image
- you want an open-weight model you can finetune on your own data
- you have a modern NVIDIA GPU (H100 or RTX 4090) for inference

## When to avoid
- you need real-time generation on modest hardware
- you only need simple lip-sync on existing video rather than generating video from a still image
- you lack a CUDA GPU

## Facets
- artifact type: library
- maturity: active
- function: video-processing, deep-learning, machine-learning, speech-recognition, llm-inference
- domain: deep-learning, computer-vision, artificial-intelligence
- platform: python, cli
- tags: talking-head-generation, diffusion-models, audio-driven-animation, portrait-animation, lip-sync, digital-avatar, research-model, video, audio, gpu, linux

## Member repositories
- memoavatar/memo (main) score 40

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:27.897377+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 2026-08-30T06:54:03.506227+00:00, confidence not recorded.
  - readme: https://github.com/memoavatar/memo (fetched 2026-08-28T04:03:27.897377+00:00, sha eae0783b7d80)
  - homepage: https://memoavatar.github.io (fetched 2026-08-29T12:56:31.708489+00:00, sha 0a25fa3770f5)
- Data as of 2026-08-30T08:39:29.467469+00:00.