xfyun-voiceclone

Voice cloning from an audio recording requires uploading a WAV, MP3, M4A, or PCM file to create a training task, polling for completion, and using the resulting voice resource ID to synthesize speech. This workflow automates auth, multipart uploads, and WebSocket TTS plumbing.

WebSocket text to speech synthesis connects to iFlytek TTS services to synthesize text into MP3, PCM, speex, or opus outputs. It uses pure-stdlib Python scripts with MD5 and HMAC signing for authentication and allows adjusting speed, volume, pitch, and sample rate.

Yes, training a custom TTS voice requires iFlytek app credentials. You must set XFYUN_APP_ID, XFYUN_API_KEY, and XFYUN_API_SECRET in your environment variables to authenticate training tasks, audio uploads, and WebSocket synthesis requests.

Yes, you can build a personalized voice assistant using pure-stdlib Python scripts. This Skill handles end-to-end TTS training and synthesis workflows without external dependencies, supporting demo audio generation and localized narration from speaker recordings.

Supported audio formats for uploading voice training samples include WAV, MP3, M4A, and PCM files. You can upload audio from a local file or via a URL to create the training task for your custom voice resource.

Pure-stdlib Python scripts for voice cloning rely entirely on iFlytek WebSocket TTS infrastructure and credentials. You must manage training task polling and handle multipart audio uploads manually within the script's workflow to obtain the synthesized speech outputs.