Voice AI app

STT (Whisper) + LLM + TTS (ElevenLabs). Push-to-talk in, AI voice out. ~60 minutes.

A voice interface: user speaks; Whisper transcribes; an LLM responds; ElevenLabs speaks the answer.

Time: ~60 min. Difficulty: Advanced. Integrations: OpenAI (Whisper, GPT-4o), ElevenLabs.

Full walkthrough being written.

Foundation prompt

A voice AI assistant. Push-to-talk button: hold to record, release to send.
Whisper transcribes the audio. The transcription goes to GPT-4o; the response
streams back. ElevenLabs converts the response to speech and plays it.

Conversation history preserved per user. Show transcripts of both user and AI.

Related

OpenAI

Whisper + GPT-4o.

ElevenLabs

TTS.

AI chatbot recipe

The text variant.