How to synthesize speech with Qwen3-TTS and build word-level alignment data with Qwen3-ForcedAligner.
