3 ms·

Went looking for audio samples, here's some from one of the researchers: http://www.zhizheng.org/demo/is15_mte/demo.html http://www.zhizheng.org/demo/is15_mte/

• raverbashing 11y ago
  Also here http://homepages.inf.ed.ac.uk/zwu2/demo/icassp16/lstm.html http://homepages.inf.ed.ac.uk/zwu2/demo/icassp16/lstm.html
  • 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.