Music & Audio Generation
Audio AI in the Wild
7 min+70 XPStep 1/10
The voice stack you already use
The pieces from this track are already stacked in production:
- Call centers — ASR transcribes every call; LLMs summarize, score, and search them
- Captions — live subtitling for meetings and streams, including fully on-device captioning for accessibility
- Voice notes — record, transcribe, summarize; it's why your messaging app can read a voice memo for you
The pattern everywhere: ASR turns speech into text, an LLM does the thinking, TTS speaks back if needed.
insight
ASR + LLM is the modern voice stack: hearing outsourced to a speech model, thinking to a language model — and each half can be upgraded independently.