Speech synthesis system and method based on rhythm emotion migration
The invention discloses a speech synthesis system and method based on rhythm emotion migration, and the system comprises a text encoder module, a sequence alignment module, a multi-level style adapter, a content adapter module, and a decoder module, the text encoder module is used for carrying out v...
Gespeichert in:
Hauptverfasser: | , , |
---|---|
Format: | Patent |
Sprache: | chi ; eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The invention discloses a speech synthesis system and method based on rhythm emotion migration, and the system comprises a text encoder module, a sequence alignment module, a multi-level style adapter, a content adapter module, and a decoder module, the text encoder module is used for carrying out vectorization coding on a text input in a TTS system, and the sequence alignment module is used for carrying out sequence alignment on the multi-level style adapter; the codes are mixed with some style attributes; the sequence alignment module is used for voice-text alignment, and the style attribute is eliminated through the content adapter module after alignment; the multi-level style adapter extracts multi-scale features of the reference audio, fuses the multi-scale features, and inputs the fused multi-scale features and the output after content adaptation into the voice frame decoder for outputting the Mel sound spectrum; and finally, accessing a vocoder to convert the Mel sound spectrum into a voice waveform. A |
---|