===== English | 简体中文 查看demo视频 Using ChatGPT to connect with FreeSWITCH, creating an intelligent phone robot.
freeswitch-chatGPT freeswitch-chatGPT is an open-source project aimed at integrating FreeSWITCH with OpenAI's Stream API, as well as implementing MRCP-based ASR (Automatic Speech Recognition) and TTS (Text-to-Speech) using FreeSWITCH.
open ai
Installation Before installing freeswitch-chatGPT, make sure you have the following prerequisites:
FreeSWITCH 1.10 or later installed and configured Python 3.6 or later installed A valid OpenAI API key To install freeswitch-chatGPT, follow these steps:
Clone the repository: bash
Copy code
git clone https://github.com/laoyin/freeswitch_chatGPT.git
Install the required Python dependencies:
Copy code
pip install -r requirements.txt
Set up your OpenAI API authentication credentials:
javascript
Copy code
export OPENAI_API_KEY=your_api_key_here
Run the installation script:
bash
Copy code
./install.sh
This will create a freeswitch-chatGPT directory in your FreeSWITCH scripts directory, and copy the necessary files there.
Usage freeswitch-chatGPT is divided into three stages, each corresponding to one of the project goals stated above. To use each stage, follow these instructions:
Stage 1: OpenAI Stream integration To use the OpenAI Stream integration, start a FreeSWITCH session and load the openai_stream.lua script. Then, call the openai_stream() function with the appropriate arguments:
lua Copy code openai_stream(model, prompt, length) where model is the name of the OpenAI model to use, prompt is the initial prompt text to send to the model, and length is the length of the generated text.
For example:
lua Copy code
openai_stream("text-davinci-002", "Hello, world!", 50)
This will generate 50 characters of text using the specified OpenAI model, starting with the "Hello, world!" prompt.
Stage 2: MRCP integration To use the MRCP integration, start a FreeSWITCH session and load the mrcp.lua script. Then, call the mrcp_asr() or mrcp_tts() function with the appropriate arguments:
lua Copy code
mrcp_asr(engine, prompt, callback_url)
lua Copy code
mrcp_tts(engine, text, callback_url)
where engine is the name of the MRCP engine to use, prompt or text are the speech or text input for the ASR or TTS operation respectively, and callback_url is the URL to which the MRCP server will send the recognition or synthesized audio data.
For example:
lua Copy code mrcp_asr("google", "Say something...", "http://myserver/callback") This will perform an ASR operation using the Google MRCP engine, prompting the user to say something, and sending the recognition result to the specified callback URL.
Stage 3: ASR and TTS To use the ASR and TTS functionality, start a FreeSWITCH session and load the asr.lua or tts.lua script. Then, call the asr() or tts() function with the appropriate arguments:
lua Copy code
asr(engine, path, lang, grammar, callback_url)
lua Copy code
tts(engine, text, path, lang, callback_url)
where engine is the name of the ASR or TTS engine to use, path is the path to the audio file to use for ASR, lang is the language code for the input or output text, grammar is the grammar file for the ASR engine, if applicable, and callback_url is the URL to which the recognition or synthesized audio data will be sent.
For example:
lua Copy code
asr("google", "/path/to/audio.wav", "en-US", "/path/to/grammar.xml", "http://myserver/callback")
This will perform ASR on the specified audio file using the Google ASR engine, with the specified grammar file and language, and sending the recognition result to the specified callback URL.
License freeswitch-chatGPT is released under the MIT License. See the LICENSE file for details.
Contributing Contributions are welcome! Please see the CONTRIBUTING file for guidelines.
Contact For questions or feedback, please contact me at [email protected]
欢迎打赏-咨询
