Python Deep Learning - Second Edition

Decoding

The process of obtaining the actual words from the audio features and phoneme distribution is known as decoding. Once we model the phoneme distribution with the acoustic model and train a language model, we can combine it with a pronunciation dictionary to get a probability function of words over audio features:

P (words | audio features) = P (audio features | phonemes) * P (phonemes | words) * P (words)

This probability function doesn't give us the final transcript yet; we still need to perform a search over the distribution of the word sequence to find the most likely transcription. This search process is called decoding. All possible paths of decoding can be illustrated in a lattice data structure (source: https://www.isip.piconepress.com/projects/speech/software/legacy/lattice_tools/ ...