5 ms·

I thought this would be about text-to-speech applications, while this seems more like an encoder-decoder problem (make the network learn a pattern and then let

by sawwit 11y ago

I thought this would be about text-to-speech applications, while this seems more like an encoder-decoder problem (make the network learn a pattern and then let it reproduce it). I'm wondering how long it is until we see working TTS based on LSTM RNNs.

ghayes 11y ago
Yeah, can someone explain the exact problem of "statistical parametric speech synthesis," since I can't find a general overview of the problem itself.
- amelius 11y ago
  I'm a newbie to all this, but I can imagine it could be useful for speech compression.