mlsanityData to 14 Oct 2026

Gender Bias in Instruction-Guided Speech Synthesis Models

arXiv 2502.05649 ↗First posted 8 Feb 2025cs.CL cs.LG eess.AScited by 4

Not on the map. Why?

Abstract (arXiv)

Recent advancements in controllable expressive speech synthesis, especially in text-to-speech (TTS) models, have allowed for the generation of speech with specific styles guided by textual descriptions, known as style prompts. While this development enhances the flexibility and naturalness of synthesized speech, there remains a significant gap in understanding how these models handle vague or abstract style prompts. This study investigates the potential gender bias in how models interpret occupation-related prompts, specifically examining their responses to instructions like "Act like a nurse". We explore whether these models exhibit tendencies to amplify gender stereotypes when interpreting such prompts. Our experimental results reveal the model's tendency to exhibit gender bias for certain occupations. Moreover, models of different sizes show varying degrees of this bias across these occupations.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 21 of its 43 references that are in the map. First posted on arXiv 8 Feb 2025.

Builds on

21 of 43 references.

References · 21

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2022Robust Speech Recognition via Large-Scale Weak Supervisioncited by 4,057
  3. 2020wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representationscited by 3,905
  4. 2019Common Voice: A Massively-Multilingual Speech Corpuscited by 1,384
  5. 2018VoxCeleb2: Deep Speaker Recognitioncited by 875
  6. 2018Gender Bias in Coreference Resolution: Evaluation and Debiasing Methodscited by 659
  7. 2020MLS: A Large-Scale Multilingual Dataset for Speech Researchcited by 476
  8. 2024CosyVoice: A Scalable Multilingual Zero-shot Text-to-speech Synthesizer based on Supervised Semantic Tokenscited by 334
  9. 2023emotion2vec: Self-Supervised Pre-Training for Speech Emotion Representationcited by 205
  10. 2023LibriTTS-R: A Restored Multi-Speaker Text-to-Speech Corpuscited by 101
  11. 2023EXPRESSO: A Benchmark and Analysis of Discrete Expressive Speech Resynthesiscited by 98
  12. 2022How well can Text-to-Image Generative Models understand Ethical Natural Language Interventions?cited by 92
9 more
  1. 2023InstructTTS: Modelling Expressive TTS in Discrete Latent Space with Natural Language Style Promptcited by 92
  2. 2024Natural language guidance of high-fidelity text-to-speech with synthetic annotationscited by 67
  3. 2023TextrolSpeech: A Text Style Control Speech Corpus With Codec Language Text-to-Speech Modelscited by 48
  4. 2023PromptTTS 2: Describing and Generating Voices with Text Promptcited by 48
  5. 2023PromptStyle: Controllable Style Transfer for Text-to-Speech with Natural Language Descriptionscited by 34
  6. 2023PromptTTS++: Controlling Speaker Identity in Prompt-Based Text-to-Speech Using Natural Language Descriptionscited by 25
  7. 2023Speech-based Age and Gender Prediction with Transformerscited by 14
  8. 2023Towards General-Purpose Text-Instruction-Guided Voice Conversioncited by 6
  9. 2024HybridVC: Efficient Voice Style Conversion with Text and Audio Promptscited by 3

Built on by

Cited by year

Citing papers per year: 2025: 4.

4 papers.

  1. 2025Evaluating Bias in Spoken Dialogue LLMs for Real-World Decisions and Recommendationscited by 7
  2. 2025Speak Your Mind: The Speech Continuation Task as a Probe of Voice-Based Model Biascited by 0
  3. 2025CLARITY: Contextual Linguistic Adaptation and Accent Retrieval for Dual-Bias Mitigation in Text-to-Speech Generationcited by 0
  4. 2025Gender Bias in Emotion Recognition by Large Language Modelscited by 0