Method and system for processing speech signal
AssigneeAlibaba Group Holding Limited
InventorsShiliang Zhang, Ming Lei, Wei Li, Haitao Yao
Life of this patent
Estimating the term…
Abstract
Embodiments of the present disclosure provide methods and systems for processing a speech signal. The method can include: processing the speech signal to generate a plurality of speech frames; generating a first number of acoustic features based on the plurality of speech frames using a frame shift at a given frequency; and generating a second number of posteriori probability vectors based on the first number of acoustic features using an acoustic model, wherein each of the posteriori probability vectors comprises probabilities of the acoustic features corresponding to a plurality of modeling units, respectively.
Claims20
A method for processing a speech signal, comprising:
- obtaining a speech signal;
- generating, using the speech signal, a first sequence of feature vectors;
- selecting, from the first sequence of feature vectors, a second sequence of m*n consecutive feature vectors;
- generating a third sequence of n intermediate vectors by applying the second sequence to a first neural network model, each intermediate vector corresponding to a subsequence of m consecutive feature vectors in the first sequence;
- generating a fourth sequence of n average probability vectors by applying each of the n intermediate vectors to a corresponding second neural network model;
- and
- determining the phones in the second sequence using the fourth sequence of n average probability vectors.
A system for processing a speech signal, comprising:
- at least one processor;
- and
- at least one non-transitory, computer-readable medium containing instructions that, when executed by the at least one processor, cause the system to perform operations comprising:
- obtaining a speech signal;
- generating, using the speech signal, a first sequence of feature vectors;
- selecting, from the first sequence of feature vectors, a second sequence of m*n consecutive feature vectors;
- generating a third sequence of n intermediate vectors by applying the second sequence to a first neural network model, each intermediate vector corresponding to a subsequence of m consecutive feature vectors in the first sequence;
- generating a fourth sequence of n average probability vectors by applying each of the n intermediate vectors to a corresponding second neural network model;
- and
- determining the phones in the second sequence using the fourth sequence of n average probability vectors.
A non-transitory, computer-readable medium containing instructions that, when executed by at least one processor of a system for processing a speech signal, cause the system to perform operations comprising:
- obtaining a speech signal;
- generating, using the speech signal, a first sequence of feature vectors;
- selecting, from the first sequence of feature vectors, a second sequence of m*n consecutive feature vectors;
- generating a third sequence of n intermediate vectors by applying the second sequence to a first neural network model, each intermediate vector corresponding to a subsequence of m consecutive feature vectors in the first sequence;
- generating a fourth sequence of n average probability vectors by applying each of the n intermediate vectors to a corresponding second neural network model;
- and
- determining the phones in the second sequence using the fourth sequence of n average probability vectors.
Dependencies are inferred from the claim text.