Being consistent is really difficult , but the payoff is HUGE!
Hereโs what works.
Started small (10 min/day) & found my groove with batching & an accountability buddy.
Now it's my secret weapon!
Joined SPRING Labs , Department of Electrical Engineering, IIT Madras as a Research Intern.
Will work broadly on ASR , TTS and Speech-2-Speech Translation (S2ST).
#Artificiallntelligence#technology
Expecting to see a surge in GenAI use in smartphones in second half of 2024.
What kind of GenAI features would you like to see on smartphones?
#GenAI#GitHub#Artificiallntelligence
Was going through HuBERT research paper ! Itโs really impressive that HuBERT can understand speech without relying on text data! This self-supervised model learns directly from audio & achieves significant WER improvements.
Finding the right speaker ID embedding distance threshold is key! Too high & you miss matches, too low & imposters sneak.
It's all about finding the sweet spot between accuracy & false positives!
Leveraging the power of speechbrain/voxceleb-ecapa-TDNN for speaker diarization! This impressive toolkit excels at extracting speaker embeddings using ECAPA-TDNN, a novel architecture combining convolutional & residual blocks for robust performance.
It's frustrating when a team member puts in a great effort but the team as a whole doesn't achieve the desired outcome. This can happen for various reasons, such as lack of coordination, communication, or overall team strategy.
#RRvsRCB#ViratKohli
@kripto1856453 Interesting prediction! The SEC lawsuit is definitely a wild card. Personally, I'm bullish long-term on XRP's utility, but who knows what the next quarter holds? $XRP