# MOSS-VoiceGenerator: Create Realistic Voices with Natural Language Descriptions

arXiv: https://arxiv.org/abs/2603.28086 · HTML: https://mlsanity.com/paper/2603.28086/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 25 references that are in the map.
- First posted on arXiv 30 Mar 2026.

## Builds on (most cited first, 22 of 22)

- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [OpenAI GPT-5 System Card](https://mlsanity.com/paper/2601.03267/) (2025)
- [Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers](https://mlsanity.com/paper/2301.02111/) (2023)
- [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://mlsanity.com/paper/2211.06687/) (2022)
- [Qwen3-Omni Technical Report](https://mlsanity.com/paper/2509.17765/) (2025)
- [Seed-TTS: A Family of High-Quality Versatile Speech Generation Models](https://mlsanity.com/paper/2406.02430/) (2024)
- [CosyVoice 2: Scalable Streaming Speech Synthesis with Large Language Models](https://mlsanity.com/paper/2412.10117/) (2024)
- [DNSMOS: A Non-Intrusive Perceptual Objective Speech Quality metric to evaluate Noise Suppressors](https://mlsanity.com/paper/2010.15258/) (2020)
- [Audiobox: Unified Audio Generation with Natural Language Prompts](https://mlsanity.com/paper/2312.15821/) (2023)
- [Qwen3-TTS Technical Report](https://mlsanity.com/paper/2601.15621/) (2026)
- [InstructTTS: Modelling Expressive TTS in Discrete Latent Space with Natural Language Style Prompt](https://mlsanity.com/paper/2301.13662/) (2023)
- [IndexTTS2: A Breakthrough in Emotionally Expressive and Duration-Controlled Auto-Regressive Zero-Shot Text-to-Speech](https://mlsanity.com/paper/2506.21619/) (2025)
- [MiMo-Audio: Audio Language Models are Few-Shot Learners](https://mlsanity.com/paper/2512.23808/) (2025)
- [EmoVoice: LLM-based Emotional Text-To-Speech Model with Freestyle Text Prompting](https://mlsanity.com/paper/2504.12867/) (2025)
- [Leveraging Self-Supervised Learning for Speaker Diarization](https://mlsanity.com/paper/2409.09408/) (2024)
- [VoxInstruct: Expressive Human Instruction-to-Speech Generation with Unified Multilingual Codec Language Modelling](https://mlsanity.com/paper/2408.15676/) (2024)
- [InstructTTSEval: Benchmarking Complex Natural-Language Instruction Following in Text-to-Speech Systems](https://mlsanity.com/paper/2506.16381/) (2025)
- [Fine-tune Before Structured Pruning: Towards Compact and Accurate Self-Supervised Models for Speaker Diarization](https://mlsanity.com/paper/2505.24111/) (2025)
- [VoiceSculptor: Your Voice, Designed By You](https://mlsanity.com/paper/2601.10629/) (2026)
- [Efficient and Robust Speaker Diarization via Structured Pruning of Self-Supervised Models](https://mlsanity.com/paper/2506.18623/) (2025)
- [MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models](https://mlsanity.com/paper/2602.10934/) (2026)

## Built on by (most cited first, 3 of 3)

- [FireRedAudio: A General-Purpose Audio Language Model with Decoupled Continuous Representations for Understanding and Generation](https://mlsanity.com/paper/2608.24168/) (2026)
- [AuK Technical Report: An Open-Source Foundational Model for Speech Generation and Editing](https://mlsanity.com/paper/2609.08936/) (2026)
- [JoyAI-Voice 2.0: A Full-Continuous Autoregressive Speech Generation Model with Semantic-Acoustic Joint Representation](https://mlsanity.com/paper/2610.08834/) (2026)

## Abstract (arXiv)

Voice design from natural language aims to generate speaker timbres directly from free-form textual descriptions, allowing users to create voices tailored to specific roles, personalities, and emotions. Such controllable voice creation benefits a wide range of downstream applications-including storytelling, game dubbing, role-play agents, and conversational assistants, making it a significant task for modern Text-to-Speech models. However, existing models are largely trained on carefully recorded studio data, which produces speech that is clean and well-articulated, yet lacks the lived-in qualities of real human voices. To address these limitations, we present MOSS-VoiceGenerator, an open-source instruction-driven voice generation model that creates new timbres directly from natural language prompts. Motivated by the hypothesis that exposure to real-world acoustic variation produces more perceptually natural voices, we train on large-scale expressive speech data sourced from cinematic content. Subjective preference studies demonstrate its superiority in overall performance, instruction-following, and naturalness compared to other voice design models.