Dataset Card for TTS SODA

This is a TTS synthesized version of SODA using ChatTTS.

Dataset Details

Dataset Information

• Language: English
• License: CC BY: Creative Commons Attribution
• Repository: TBA
• Paper: TBA

Data Fields

• input_audio: TTS synthesized audio of the input_transcript
• output_audio: TTS synthesized audio of the output_transcript
• original_index: Index of dialogue from the original SODA dataset
• dialogue_turn: Dialogue turn number of the input_audio
• input_tts_wer: WER of Whisper V3 transcript of input_audio against the input_transcript
• output_tts_wer: WER of Whisper V3 transcript of output_audio against the output_transcript
• input_tokens: Speechtokenizer tokens of the input_audio
• output_tokens: Speechtokenizer tokens of the output_audio
• input_transcript: Text of input query
• output_transcript: Text of output response

Citation

BibTeX:

TBA