SpeechGym: An Audio-Native Gym for Training Voice Agents via Reinforcement Learning

arXiv:2608.26432 · cs.SD, cs.AI, cs.CL · Submitted 2026-08-26 · Read on arXiv

cs.SD, cs.AI, cs.CL

Submitted: 2026-08-26

Updated: 2026-08-26

Code: https://github.com/shuaijiang/Ke-Omni-R

Terminology

• Benchmark — 17× in this paper · explained in 2 episodes
• GRPO — 12× in this paper · explained in 6 episodes
• Agentic Task — 6× in this paper · explained in 1 episode
• Reinforcement Learning — 6× in this paper · explained in 4 episodes
• Agentic Tasks — 5× in this paper · explained in 1 episode
• Benchmarking — 3× in this paper · explained in 3 episodes
• Pruning — 3× in this paper · explained in 4 episodes
• Agentic Loop — 2× in this paper · explained in 1 episode
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Generative Model — 2× in this paper · explained in 2 episodes
• Generative Models — 2× in this paper · explained in 2 episodes
• Reward Shaping — 2× in this paper · explained in 2 episodes
• Adaptation — 1× in this paper · explained in 1 episode
• Calibration — 1× in this paper · explained in 7 episodes
• Credit assignment — 1× in this paper · explained in 3 episodes
• Dec-POMDP — 1× in this paper · explained in 2 episodes
• Failure Modes — 1× in this paper · explained in 3 episodes
• Flow Matching — 1× in this paper · explained in 8 episodes
• Foundation Model — 1× in this paper · explained in 3 episodes
• Gradient Flow — 1× in this paper · explained in 2 episodes
• Multi-hop Reasoning — 1× in this paper · explained in 3 episodes
• Regularization — 1× in this paper · explained in 3 episodes
• Reward Hacking — 1× in this paper · explained in 8 episodes
• Sparsity — 1× in this paper · explained in 4 episodes
• Trace — 1× in this paper · explained in 2 episodes
• Verifiable Rewards — 1× in this paper · explained in 2 episodes

Sources

• $\tau^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment
• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
• Moshi: a speech-text foundation model for real-time dialogue
• Incentivizing Consistent, Effective and Scalable Reasoning Capability in Audio LLMs via Reasoning Process Rewards
• VITA: Towards Open-Source Interactive Omni Multimodal LLM
• VoiceAgentBench: Are Voice Assistants ready for agentic tasks?
• ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models
• Full-Duplex-Bench-v3: Benchmarking Tool Use for Full-Duplex Voice Agents Under Real-World Disfluency
• Multi-Turn Reinforcement Learning for Tool-Calling Agents with Iterative Reward Calibration
• $\tau$-Voice: Benchmarking Full-Duplex Voice Agents on Real-World Domains
• Proximal Policy Optimization Algorithms
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Retrieval is Cheap, Show Me the Code: Executable Multi-Hop Reasoning for Retrieval-Augmented Generation
• Qwen3-Omni Technical Report
• Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming
• Qwen2.5-Omni Technical Report
• $\tau$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains

Related papers

• Few-Shot Open-Set Audio Classification via Transductive Prototype Refinement and Class Logit Enhancement
• Spectral Masking and Interpolation Attack (SMIA): A Black-box Adversarial Attack against Voice Authentication and Anti-Spoofing Systems
• AVMeme Exam: A Multimodal Multilingual Multicultural Benchmark for LLMs' Contextual and Cultural Knowledge and Thinking
• SoundWeaver: Compositional Warm-Starting for Text-to-Audio Diffusion Serving
• WASIL: In-the-Wild Arabic Spoken Interactions with LLMs
• Efficient Audiovisual Speech Processing via MUTUD: Multimodal Training and Unimodal Deployment