urdu-tts-corpus

Urdu TTS Corpus 是一个精心整理的乌尔都语语音-文本配对数据集，专为训练文本到语音（TTS）和自动语音识别（ASR）模型而设计。该数据集将多个高质量来源整合为标准化的格式。数据集包含以下特征：hash_id（字符串）、text（字符串）、audio（音频，采样率为16,000 Hz）、duration_ms（整数）和src（字符串）。数据集分为训练集（train），包含122,477个样本，总大小为9,333,384,310字节。数据集适用于文本到语音和文本到音频任务，语言为乌尔都语（ur-PK），采样率为16,000 Hz，格式为Hugging Face数据集（音频+文本）。数据集合并了四个来源：gondal_urdu_tts、urdu_tts_16k、mozilla_cv_urdu_24和urdu_tts_fast。

Urdu TTS Corpus is a meticulously curated Urdu speech-text paired dataset specifically designed for training text-to-speech (TTS) and automatic speech recognition (ASR) models. This dataset integrates multiple high-quality sources into a standardized format. The dataset includes the following features: hash_id (string), text (string), audio (audio with a sampling rate of 16,000 Hz), duration_ms (integer), and src (string). The dataset is split into the training set (train), which contains 122,477 samples with a total size of 9,333,384,310 bytes. It is suitable for text-to-speech and text-to-audio tasks, in Urdu (ur-PK), with a sampling rate of 16,000 Hz, and formatted as a Hugging Face dataset (audio + text). The dataset combines four sources: gondal_urdu_tts, urdu_tts_16k, mozilla_cv_urdu_24, and urdu_tts_fast.