We're currently building an application based in the Facebook Wav2Letter speech to text engine and require an experienced NTL specialist / data scientist to assist us with the creation of the necessary data to train the models we require for the engine.
The tool will have multiple versions, each with a specific model that is optimized for a specific niche or accent, for example UK Legal, Medical etc. We plan to start by using an open-source model(s) such as librispeech / common voice etc to form a foundation but then wish to add specialized vocabulary to target a very low WER rate in our chosen niches.
We have a number of options available for audio files, dictionaries etc and want to work with a highly experience data scientist, NLP expert to find the best way to develop these specialized models and then be able to take the data and train the models.
About the recuiterMember since May 20, 2018 Rajesh N
from Wiltz, Luxembourg