Voice identification for optimizing voice search results

A system for processing a voice input stream comprising a first input and a second input, the system comprising:

• input/output circuitry configured to:
• receive, at a microphone of an electronic device, the first input comprising a voice query from a first voice;
• receive the second input comprising a secondary query from a second voice;
• and
• processing circuitry configured to:
• stream, from the electronic device to a cloud service, the received first and second voice inputs for analysis, wherein the analysis by the cloud service comprises:
• inputting the received first and second voice inputs into a machine learning model trained to determine a sound match score and a substance match score, wherein:
• the sound match score comprises a probability that voices corresponding to two separate voice inputs match;
• the substance match score comprises a probability that the substance of two separate voice inputs match;
• determining a weighted average of the sound match score and the substance match score;
• and
• comparing the weighted average of the sound match score and the substance match score to a confidence threshold;
• determine, based on the analysis by the cloud service, that the second voice does not match the first voice;
• and
• in response to determining that the second voice does not match the first voice:
• process the voice query, without the second query;
• execute a first search, by the electronic device, to obtain first results that correspond to the voice query from the first voice;
• filter the obtained first results using one or more keywords from the secondary query from the second voice;
• determine the obtained number of first results do not exceed a threshold number of first results;
• based at least in part on determining the obtained number of first results do not exceed a threshold number of first results, automatically execute a second search, by the electronic device, to obtain second results that correspond to the voice query from the first voice and the second voice;
• and
• produce the filtered first results and at least one second result, wherein the second result is based at least in part on one or more keywords from second voice, via a speaker or display associated with the electronic device.