Alibaba released Qwen-Audio-3.1 at its Apsara conference: five models covering speech recognition, text to speech and realtime voice, plus TTS-Next for generation and ASR-Next for understanding. It cut the prices with them. Speech recognition falls by up to 95%, text to speech by about 70%, realtime voice by about 85%. ASR-Next adds speaker identification with timestamps, emotion detection and ambient sound. TTS-Next makes voice, effects and background audio in one pass. Alibaba has not published the new rates in dollars, so the cuts are recorded here as the percentages it announced.