Audio processing method and related apparatus

An audio processing apparatus, comprising:

• at least one memory configured to store computer program code;
• and
• at least one processor configured to read the program code and operate as instructed by the program code, the program code comprising:
• audio signal acquiring code configured to cause at least one of the at least one processor to acquire an audio signal comprising one or more audio frames;
• streaming acoustic network processing code configured to cause at least one of the at least one processor to input the one or more audio frames into a streaming acoustic network to obtain one or more phoneme features representing phoneme information of the audio signal and one or more streaming audio features;
• entity set acquiring code configured to cause at least one of the at least one processor to acquire an entity set comprising one or more first entities, wherein the one or more first entities correspond to one or more pieces of phoneme information;
• entity extraction code configured to cause at least one of the at least one processor to extract one or more second entities from the entity set based on the one or more phoneme features, wherein the one or more second entities correspond to the one or more phoneme features, and wherein a second number of the one or more second entities is greater than or equal to a third number of the one or more audio frames and less than or equal to a first number of the one or more first entities;
• non-streaming acoustic network processing code configured to cause at least one of the at least one processor to obtain a text recognition result based on inputting the audio signal, the one or more streaming audio features, and the one or more second entities into a non-streaming acoustic network;
• and
• first outputting code configured to cause at least one of the at least one processor to output the text recognition result.

A non-transitory computer-readable storage medium, storing computer code which, when executed by at least one processor, causes the at least one processor to at least:

• acquire an audio signal comprising one or more audio frames;
• input the one or more audio frames into a streaming acoustic network to obtain one or more phoneme features representing phoneme information of the audio signal and one or more streaming audio features;
• acquire an entity set comprising one or more first entities, wherein the one or more first entities correspond to one or more pieces of phoneme information;
• extract one or more second entities from the entity set based on the one or more phoneme features, wherein the one or more second entities correspond to the one or more phoneme features, and wherein a second number of the one or more second entities is greater than or equal to a third number of the one or more audio frames and less than or equal to a first number of the one or more first entities;
• obtain a text recognition result based on inputting the audio signal, the one or more streaming audio features, and the one or more second entities into a non-streaming acoustic network;
• and
• output the text recognition result.