Dear users, you've might been asking yourself why there were not so many updates on CMUSphinx recently. Time goes really fast and many things change in ASR. Deep learning, huge NLP models like BERT, Tacotron and Wavenet/Waveglow/WaveRNN, Pytorch vs Tensorflow, huge datsets, chatbots and so on and so forth. Many new toolkits appear and some disappear - Eesen, Espresso, Kaldi, Wav2letter, NeMo. The whole area is thriving.
CMUSphinx team has been actively participating in all those activities, creating new models, applications, helping newcomers and showing the best way to implement speech recognition system. We are here to suggest you the easiest way to start such an exciting world of speech recognition. Lately we implemented a Kaldi on Android, providing much better accuracy for large vocabulary decoding, which was hard to imagine before.
Hi everyone! My name’s Sahith Dambekodi and I’m a second year undergrad student at BITS Pilani K.K. Birla Goa Campus. I’m pursuing a major in Electrical and Electronics engineering. This is my blog for Google Summer of Code 2017!
The list of accepted projects for Google Summer of Code 2017 has been announced today. Please check
There are two major parts, one is pronunciation evaluation, we have several sub-projects about it, another part is about deep neural networks in pocketsphinx. Hopefully, the accuracy of our decoders will improve significantly. An interesting project is dedicated to more tight ROS integration, hopefully using speech recognition in robots will be way easier soon.
Congratulations to accepted students! Great summer ahead, lots of fun it seems.