qwen3-tts-rs Profiling & Benchmarking

Run performance profiling and benchmarks for the qwen3-tts Rust TTS engine.

Prerequisites

• Docker with --gpus all support
• qwen3-tts:latest Docker image (has Rust toolchain + CUDA)
• Model weights in test_data/models/ (1.7B-CustomVoice is the default)
• tokenizer.json must be in the model directory

Docker Execution Pattern

The CUDA toolchain lives inside the Docker container. All cargo commands must run there. The workspace is bind-mounted at /workspace: