Building a Python Voice Synthesis System Using Xunfei WebAPI

Background: The author is interested in voice synthesis and wants to build a tool that can read e‑books aloud.

System Overview: By leveraging Xunfei Open Platform's WebAPI, a lightweight voice synthesis system is constructed, reducing development complexity to a few API calls.

Prerequisites: Install Anaconda, Python 3.7, and Visual Studio Code on the development machine.

Steps: Create an application on the Xunfei console, obtain APPID, APIKey, and APISecret, then install the required Python libraries with pip install websocket-client and pip install playsound .

Implementation: Define a class play containing initialization, play_sound , select_vcn , and xfyun_tts methods; set up a Tkinter GUI with text box, voice‑selection combobox, and control buttons.

Core TTS Logic: The Ws_Param class configures business and common arguments, generates an HMAC‑SHA256 signature, builds the authenticated WebSocket URL, and provides callbacks ( on_message , on_error , on_close , on_open ) to handle streaming audio data, saving PCM files and converting them to WAV.

Final Result: The completed application allows users to input text, choose from several voice profiles, synthesize speech via Xunfei's service, and play the generated audio directly from the GUI.