# Adding Voice Cloning to Text-to-Audio-Video Models with a Single Zero-Initialised Layer

arXiv: https://arxiv.org/abs/2608.15690 · HTML: https://mlsanity.com/paper/2608.15690/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 19 references that are in the map.
- First posted on arXiv 16 Aug 2026.

## Builds on (most cited first, 17 of 17)

- [Classifier-Free Diffusion Guidance](https://mlsanity.com/paper/2207.12598/) (2022)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [HunyuanVideo: A Systematic Framework For Large Video Generative Models](https://mlsanity.com/paper/2412.03603/) (2024)
- [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://mlsanity.com/paper/2110.13900/) (2021)
- [Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers](https://mlsanity.com/paper/2301.02111/) (2023)
- [ECAPA-TDNN: Emphasized Channel Attention, Propagation and Aggregation in TDNN Based Speaker Verification](https://mlsanity.com/paper/2005.07143/) (2020)
- [Generalized End-to-End Loss for Speaker Verification](https://mlsanity.com/paper/1710.10467/) (2017)
- [YourTTS: Towards Zero-Shot Multi-Speaker TTS and Zero-Shot Voice Conversion for everyone](https://mlsanity.com/paper/2112.02418/) (2021)
- [LTX-2: Efficient Joint Audio-Visual Foundation Model](https://mlsanity.com/paper/2601.03233/) (2026)
- [XTTS: a Massively Multilingual Zero-Shot Text-to-Speech Model](https://mlsanity.com/paper/2406.04904/) (2024)
- [MMAudio: Taming Multimodal Joint Training for High-Quality Video-to-Audio Synthesis](https://mlsanity.com/paper/2412.15322/) (2024)
- [IndexTTS2: A Breakthrough in Emotionally Expressive and Duration-Controlled Auto-Regressive Zero-Shot Text-to-Speech](https://mlsanity.com/paper/2506.21619/) (2025)
- [SyncFlow: Toward Temporally Aligned Joint Audio-Video Generation from Text](https://mlsanity.com/paper/2412.15220/) (2024)
- [Kandinsky 5.0: A Family of Foundation Models for Image and Video Generation](https://mlsanity.com/paper/2511.14993/) (2025)
- [Native Audio-Visual Alignment for Generation](https://mlsanity.com/paper/2605.30073/) (2026)
- [3MDiT: Unified Tri-Modal Diffusion Transformer for Text-Driven Synchronized Audio-Video Generation](https://mlsanity.com/paper/2511.21780/) (2025)

## Built on by (most cited first, 1 of 1)

- [Kandinsky 6.0 Video: Foundation Models for Synchronized Video and Audio Generation](https://mlsanity.com/paper/2610.05608/) (2026)

## Abstract (arXiv)

Text-to-audio-video (T2AV) generation models produce a video and its soundtrack from a textual description, but offer no control over whose voice speaks in the output. We show that a base T2AV model can be turned into a voice-cloning model by adding a single zero-initialized linear layer on top of its audio backbone, fine-tuning for a comparatively short training schedule, and conditioning on a short reference recording at inference time. The reference is injected through two complementary signals: its diffusion latents are prepended to the audio stream, and a global speaker embedding modulates token of the target audio. On a benchmark of 674 speaker-text pairs spanning 30 speakers we compare against five strong voice-cloning text-to-speech baselines: our enhanced 5B model attains the highest speaker-encoder cosine similarity (SECS) across three independent verification networks (ECAPA-TDNN, WavLM-SV, Resemblyzer), statistically significantly outperforming every baseline. A side product of the architecture is that the audio path can be evaluated without the video path at inference time, yielding a ~30x speed-up over the full audio-video diffusion loop while preserving the voice-cloning behaviour.