Dataset · google
Basic Google voice commands dataset
google/speech_commands
Tens of thousands of one-second audio recordings of basic English words and background noise for command recognition training.
What's inside
The set consists of one-second .wav files where different speakers pronounce single English words (from 30 words in v0.01 to 35 in v0.02, including commands like 'Yes', 'No', 'Stop', 'Go', and digits) or background noise is recorded. The first version contains 64,727 recordings, with a total of 100k to 1M examples in the dataset.
What you can do with it
- Training compact keyword spotting models
- Benchmarking lightweight neural networks for audio classification on mobile devices
- Training voice activity detectors accounting for background noise
- Tasks
- Audio classification
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 11K
- Likes
- 66
Models trained on it
- nvidia/nemotron-speech-streaming-en-0.6bAutomatic speech recognition 106.4K
- nvidia/parakeet-unified-en-0.6bAutomatic speech recognition 7.6K
- nvidia/nemotron-labs-unified-asr-0.6bAutomatic speech recognition 3.6K
- animaslabs/nemotron-speech-streaming-en-0.6b-mlxAutomatic speech recognition 190
- m1el/nemotron-speech-streaming-0.6B-ggufAutomatic speech recognition 120
- animaslabs/nemotron-speech-streaming-en-0.6b-mlx-8bitAutomatic speech recognition 86
- diegomponte/wav2vec2-base-speech-commands-35Audio classification 76
- animaslabs/nemotron-speech-streaming-en-0.6b-mlx-4bitAutomatic speech recognition 45
How to load it
from datasets import load_dataset
ds = load_dataset("google/speech_commands")