Speech recognition method and apparatus with cascaded hidden layers and speech segments, computer device, and computer-readable storage medium

A speech recognition apparatus, a speech recognition model being set in the apparatus, and the apparatus comprising:

• a memory storing computer program instructions;
• and a processor coupled to the memory and configured to execute the computer program instructions and perform:
• obtaining speech data;
• performing feature extraction on the speech data, to obtain speech features of at least two speech segments in the speech data;
• inputting the speech features of the at least two speech segments into the speech recognition model, and processing the speech features of the speech segments by using n cascaded hidden layers in the speech recognition model, to obtain hidden layer features of the speech segments, comprising:
• inputting a jth intermediate hidden layer feature of an ith speech segment and a jth intermediate hidden layer feature of the (i+1)th speech segment into a (j+1)th intermediate hidden layer in the speech recognition model, to output a (j+1)th intermediate hidden layer feature of the ith speech segment, the jth intermediate hidden layer feature of the (i+1)th speech segment being obtained by the jth intermediate hidden layer through an operation based on:
• a speech feature of the (i+1)th speech segment and a speech feature of an (i+2)th speech segment in response to j being 0, and based on a (j−1)th intermediate hidden layer feature of the (i+1)th speech segment and a (j−1)th intermediate hidden layer feature of the (i+2)th speech segment in response to j being a positive integer, and both n and i are positive integers;
• and
• obtaining text information corresponding to the speech data based on the hidden layer features of the speech segments.

A non-transitory computer-readable storage medium storing computer program instructions executable by at least one processor to perform:

• obtaining speech data;
• performing feature extraction on the speech data, to obtain speech features of at least two speech segments in the speech data;
• inputting the speech features of the at least two speech segments into the speech recognition model, and processing the speech features of the speech segments by using n cascaded hidden layers in the speech recognition model, to obtain hidden layer features of the speech segments, comprising:
• inputting a jth intermediate hidden layer feature of an ith speech segment and a jth intermediate hidden layer feature of the (i+1)th speech segment into a (j+1)th intermediate hidden layer in the speech recognition model, to output a (j+1)th intermediate hidden layer feature of the ith speech segment, the jth intermediate hidden layer feature of the (i+1)th speech segment being obtained by the jth intermediate hidden layer through an operation based on:
• a speech feature of the (i+1)th speech segment and a speech feature of an (i+2)th speech segment in response to j being 0, and based on a (j−1)th intermediate hidden layer feature of the (i+1)th speech segment and a (j−1)th intermediate hidden layer feature of the (i+2)th speech segment in response to j being a positive integer, and both n and i are positive integers;
• and
• obtaining text information corresponding to the speech data based on the hidden layer features of the speech segments.