Leverage Qwen (DashScope/百炼) for speech processing: (1) Transcribe user audio (e.g., Telegram .ogg opus, wav, mp3) to text using qwen3-asr-flash, optionally with coarse timestamps; (2) Generate speech from text using qwen3-tts-flash with voice selection (default: Cherry), outputting as .ogg voice notes for Telegram.