Intelligent products and systems using Automatic Speech Recognition (ASR) are on the rise.  Applications like Google Alexa or voice-enabled car navigation systems provide next-level user experiences, yet they require a variety of speech data as training to properly recognize user commands.  We help companies working in this field by providing large amounts of relevant speech data in a variety of languages or settings, for machine learning or AI purposes.

We are able to record your script in a variety of languages, speaker sizes, utterances, age ranges, and gender distributions.

Following your project parameters, we can record in a studio setting (no background noise) or non-studio setting (some background noise accepted).  Such applications also usually require speaking in a natural utterance (without voice acting), and delivering one audio file per sentence – all for the purpose of facilitating the training process of your speech recognition engine.

You can imagine how many thousands of files could be delivered at the end of such projects – and this is not a problem for us!

For other types of applications around ASR such as language detection, the collection of large amounts of text and audio corpus is needed.

This is different than above in that naturally occurring speech from non-controlled environments are preferred.  In this case, we are able to collect text and audio data following very precise parameters which can include languages, number of speakers, length requirements, subject matter type (e.g. legal, editorial, scientific research, social media…), and more.

We have been able to collect data for even the most uncommon of dialects, so contact us with your parameters and give us a try!

Looking for speech data to train your ASR system?  Contact us now to discuss your project!