3 ms·
There is a lot of success in translating speech into a foreign language speech in near real time. The next step would be to convert heavily accented speech (exa
by RNeff 7y ago
There is a lot of success in translating speech into a foreign language speech in near real time. The next step would be to convert heavily accented speech (example: English) into normal pronunciation speech (example: English). This is called "auto-correct for accents". This would make webcast conferences (example: Google IO) more comprehensible.
- TrollHammarenF 7y agoThanks for the comment. But what i'm trying to do is to figure out a topic for my master's thesis (about automatic speech recognition). i've been searching for the latest researches about automatic speech recognition (ASR). Lots of work have been using Attention and end-to-end models. But I can't figure out a topic to explore on my master's thesis.