The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.

Description

672-hour Multi-person Meeting Multi-channel Speech Dataset covers meeting scenarios with 3-6 participants, collected in various conference room environments, mirroring real-world meeting interactions. Transcribed with text content, speaker's ID, gender, location and other attributes.

For more details, please refer to the link: https://www.nexdata.ai/datasets/speechrecog/1203?source=huggingface

Specifications

Far-field 16-microphone array

48kHz, 16bit, wav, 16channels;

Far-field 8-microphone array

8kHz, 16bit, wav, 8 channels;

Far-filed high-fidelity microphone

48kHz, 16bit, wav, mono channel;

Near-field mobile phone

16kHz, 16bit, wav, mono channel.

Recording Environment

Four different-sized conference rooms, with each size specification including three different rooms.

Recording content

Simulate a real meeting scenario;

Demographics

984 Chinese;

Annotation

extract and annotate individual sentences with their start and end timestamps, speaker identification, and spoken text content;

Device

16-microphone array, 8-microphone array, high-fidelity microphone, mobile phone;

Language

mandarin;

Application scenarios

speech recognition; voiceprint recognition;

Accuracy rate

sentences accuracy rate of 97%

Licensing Information

Commercial License

Downloads last month
44