khursanirevo/multiturn_ks_-yIVMNm2ZQU

---
license: cc-by-4.0
task_categories:
- automatic-speech-recognition
- speaker-diarization
language:
- en
- ms
- zh
- ru
- id
- ar
- ja
- ko
multilinguality:
- highly_multilingual
size_categories:
- n<1K
---

# khursanirevo/multiturn_ks_-yIVMNm2ZQU

## Dataset Description

Multiturn dialogue dataset with speaker-separated stereo audio and multi-language transcripts.

### Features

- **Audio**: Stereo audio with speaker separation (speaker 0 = left channel, speaker 1 = right channel)
- **Segments**: Speaker turn-level annotations with timestamps
- **Multi-language**: Transcripts in 9 languages (en, ms, zh-Hans, zh-Hant, ru, id, ar, ja, ko)
- **Chunking**: 30-second chunks with 0.5s overlap

### Columns

- `audio`: Playable stereo audio (24kHz)
- `sentence`: Full transcript for the chunk (English)
- `segments`: JSON list of speaker turns with speaker, start, end, text fields
- `total_speakers`: Number of speakers in chunk (typically 2)
- `sentence_ms`, `sentence_en`, etc.: Transcripts in each language

### Usage

```python
from datasets import load_dataset

dataset = load_dataset("khursanirevo/multiturn_ks_-yIVMNm2ZQU")

# Access audio and segments
chunk = dataset[0]
audio = chunk["audio"] # Stereo audio array
segments = json.loads(chunk["segments"]) # Speaker turns

for seg in segments:
speaker = seg['speaker']
text = seg['text']
print(f"Speaker {speaker}: {text}")
```

### Speaker Detection

Speakers are detected using RMS energy analysis:
- Channel 0 (left): Speaker 0
- Channel 1 (right): Speaker 1

### Languages

Supported languages:
- English (en)
- Malay (ms)
- Chinese Simplified (zh-Hans)
- Chinese Traditional (zh-Hant)
- Russian (ru)
- Indonesian (id)
- Arabic (ar)
- Japanese (ja)
- Korean (ko)

## Dataset Statistics

- Total chunks: 204
- Max chunk duration: 30.0s
- Overlap: 0.5s
- Languages: en, ms

## Source

Created from YouTube video with dialogue separation using DialogueSidon model.

## License

CC-BY-4.0