MagicHub/magicdata-dialect-henan-dialect-tts-lite

1---2language:3- zh4tags:5- Dialect6- OpenSource7- TTS8---9# MagicData-Dialect-Northeastern Chinese-TTS-Lite10## MAGIC DATA OPEN-SOURCE LICENSE11 12## Dataset Overview13| Item | Information |14| ---- | ---- |15| Dataset Type | N/A |16| Language | Chinese Dialect |17| Speech Style | Scripted |18| Content | N/A |19| Audio Parameters | 48 kHz, 16 bits |20| File Format | WAV (PCM) |21| Recording Equipment | microphone |22| Recording Environment | quiet indoor environment |23| License | Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License |24 25### Statistics26| Dialect | City | Code | Duration | Sentences | Speaker |27| ---- | ---- | ---- | ---- | ---- | ---- |28| Henan Chinese | Zhengzhou | HEN | 10 minutes | 74 sentences | 1 male, 34 years old |29 30## Dataset Introduction31MagicData-Dialect-Henan Dialect-TTS-Lite is an open-source Henan Dialect TTS subset of the MagicData-Dialect-TTS-Lite collection released by Magic Data. It focuses on authentic Henan dialect speech and is designed for research scenarios such as dialect speech synthesis, acoustic analysis, and model evaluation.32 33The dataset contains approximately 10 minutes of speech data, recorded by one native Henan dialect speaker from Zhengzhou. The speaker was born and raised in the local region, and the recordings preserve authentic local accent, intonation, and expression habits.34 35## Dataset Features361. Native speaker with authentic accent37- The speaker was born and raised in the local region until adulthood.38- The speaker’s family and main social environment use the local dialect.39 402. Daily-life content coverage41- Weather, food, family conversations, numbers, time, and dates42- A small amount of emotional expression43- No complex technical terms, news reading, or poetry recitation, in order to avoid style deviation44 453. Clean recording environment46- Quiet indoor environment47- 48 kHz / 16-bit / mono WAV48 494. Moderate sentence length, suitable for TTS modeling50- Each sentence is around 5–20 seconds, with an average length of about 10 seconds51- Natural punctuation-based segmentation, with no forced truncation52 53## Annotation Guidelines541. Chinese character transcription: Standard Chinese characters are used, while dialect-specific words are preserved and restored, such as “咋整” in Northeastern Chinese.552. Number annotation: Numbers are written in Chinese character form.563. Standardization rule: The original dialect sentences are preserved and are not forcibly “translated” into Mandarin.57 58## Open-source File Structure59dialect-tts-lite/60├── 河南61│ ├── ProsodyLabeling62│ │ ├── txt63│ ├── wav64│ │ ├── wav/ # 74 audio files65 66## Usage Recommendations67### Suitable for68- Zero-shot / few-shot baseline testing for multi-dialect TTS models69- Acoustic analysis of dialectal phonetic features70- Comparative experiments in academic research71 72### Not suitable for73- Directly training production-level dialect TTS products, as the dataset is non-commercial and limited in scale74- Evaluating extreme scenarios, such as noisy environments, far-field recording, or children’s voices75 76If you are interested in a larger-scale commercial version, please contact us.77 78## Open-source License79This dataset is for non-commercial use only under the CC BY-NC-ND 4.0 license. It is suitable for academic research, personal development, and model evaluation.80 81### Business Contact82📧 business@magicdatatech.com