724: Decoding Speech from Raw Brain Activity, with Dr. David Moses
Super Data Science: ML & AI Podcast with Jon Krohn
00:00
Decoding Speech from Brain Activity
This chapter delves into the innovative process of decoding speech from raw brain activity using a bidirectional recurrent neural network. The team utilizes a collaborative effort to collect training data where sentences are shown and participants silently attempt to say them due to paralysis. The model achieves remarkable results, decoding speech at 75 words per minute with a thousand-word vocabulary and 75% accuracy.
Play episode from 13:31
Transcript
