What’s inside Wiip’s Speech Datasets?
Each entry in the dataset consists of a unique WAV and corresponding MP3 and text file. Many of the recorded hours in the dataset also include demographic metadata like age, sex, and accent that can help train the accuracy of speech recognition engines.
The dataset currently consists of 3,200 validated hours in 20 languages, but we’re always adding more voices and languages.
Create custom datasets with the Wiip Annotation tool
Our multilingual transcription team, supervised by our Speech scientists, manually build the best curated speech datasets available in the market. Our datasets guarantee that you will get the best Automatic Speech Recognition. As AI experts, we understand the requirements in terms of quality, variety and quantity of data needed to build the best models. We provide more than just data transcripts. As an AI company we understand the whole AI life cycle, and we are aware of how you will use our data.
All our transcriptions pass through different quality steps, from an initial automatic annotation to a human annotation and a final quality revision. We have developed our own optimized transcription tool in order to ensure the best quality and fastest service on the market.
You can ask for our pre-built datasets, or if you have specific requirements, we can create your own customized dataset. All our datasets contain age-gender-accent information.