This project is a simple demo that utilizes TTS-1, Whisper, and GPT-3.5-turbo for conversational interactions.
- The front end is built using
bootstarp
,vue2
, andrecorder
. - The front end supports selecting browser-based speech-to-text (low latency) or calling OpenAI's Whisper (high accuracy).
- Users can upload
mp3
format files to PHP and obtainopus
audio through TTS to ensure efficiency. - Accessing
clean.php
allows for the clearing of cache files.
- Removed Composer dependencies, and instead integrated a modified version of the OpenAI library, incorporating functions related to TTS.
- Now utilizes a custom simple function to retrieve
.env
variables.
For optimal performance, it is recommended to use PHP 8.1.
- Rename
.env.example
to.env
and configure the OpenAI Key and URL inside it (modify the URL if a proxy is required).