tricky-tts-kokoro
收藏资源简介:
该数据集包含音频生成任务的相关数据,主要特征包括:文本提示(text_prompt)、生成的音频(generated_audio,采样率24kHz)、音频时长(duration_s)、音频标记数量(num_audio_tokens)、ASR转录文本(asr_transcription)、ASR词错误率(asr_wer)和ASR字错误率(asr_cer)。数据集仅包含训练集(train),共4个样本,总大小约4.14MB。适用于音频生成、语音合成及自动语音识别(ASR)相关任务的质量评估研究。
This dataset encompasses data for audio generation tasks, with its core features including: text prompt (text_prompt), generated audio (generated_audio, sampling rate: 24 kHz), audio duration (duration_s), number of audio tokens (num_audio_tokens), ASR transcription text (asr_transcription), ASR word error rate (asr_wer), and ASR character error rate (asr_cer). The dataset only contains the training set (train), with a total of 4 samples and an approximate total size of 4.14 MB. It is suitable for quality evaluation research on audio generation, speech synthesis, and automatic speech recognition (ASR) related tasks.




