Speaker recognition

A system comprising:

• an input for receiving an audio signal representing speech;
• and
• a processor configured to
• obtain a sample of speech of the enrolled speaker;
• obtain a measure of a fundamental frequency of the speech of the enrolled speaker in each of a first plurality of speech frames of the sample of speech of the enrolled speaker;
• and
• form a first distribution function of the fundamental frequencies in each of the first plurality of speech frames of the speech of the enrolled speaker,
• (i) for a speaker to be recognised:
• obtain a sample of speech of the speaker to be recognised;
• obtain a measure of a fundamental frequency of the speech of the speaker to be recognised in each of a second plurality of speech frames of the sample of speech of the speaker to be recognised;
• and
• form a second distribution function of the fundamental frequencies in each of the second plurality of speech frames of the speech of the speaker to be recognised,
• (ii) compare the second distribution function and the first distribution function, and
• (iii) determine whether the speaker to be recognised is the enrolled speaker based on a result of comparing the second distribution function and the first distribution function.

A non-transitory storage medium having stored thereon software code which is configured to, when run on a suitable processor

• obtain a sample of speech of the enrolled speaker;
• obtain a measure of a fundamental frequency of the speech of the enrolled speaker in each of a first plurality of speech frames of the sample of speech of the enrolled speaker;
• and
• form a first distribution function of the fundamental frequencies in each of the first plurality of speech frames of the speech of the enrolled speaker,
• (i) for a speaker to be recognised:
• obtain a sample of speech of the speaker to be recognised;
• obtain a measure of a fundamental frequency of the speech of the speaker to be recognised in each of a second plurality of speech frames of the sample of speech of the speaker to be recognised;
• and
• form a second distribution function of the fundamental frequencies in each of the second plurality of speech frames of the speech of the speaker to be recognised,
• (ii) compare the second distribution function and the first distribution function, and
• (iii) determine whether the speaker to be recognised is the enrolled speaker based on a result of comparing the second distribution function and the first distribution function.

A method of speaker change detection, comprising:

• obtaining a sample of speech;
• in each of a plurality of time periods, obtaining a measure of a fundamental frequency of the speech;
• comparing the measure of the fundamental frequency of the speech with previously obtained measures of the fundamental frequency of the speech;
• and
• determining that a speaker may have changed in the event that the measure of the fundamental frequency of the speech differs from previously obtained measures of the fundamental frequency of the speech.

A system comprising:

• an input for receiving an audio signal representing speech;
• and
• a processor configured to
• obtain a sample of speech;
• in each of a plurality of time periods, obtain a measure of a fundamental frequency of the speech;
• compare the measure of the fundamental frequency of the speech with previously obtained measures of the fundamental frequency of the speech;
• and
• determine that a speaker may have changed in the event that the measure of the fundamental frequency of the speech differs from previously obtained measures of the fundamental frequency of the speech.

A non-transitory storage medium having stored thereon software code which is configured to, when run on a suitable processor, obtain a sample of speech;

• in each of a plurality of time periods, obtain a measure of a fundamental frequency of the speech;
• compare the measure of the fundamental frequency of the speech with previously obtained measures of the fundamental frequency of the speech;
• and
• determine that a speaker may have changed in the event that the measure of the fundamental frequency of the speech differs from previously obtained measures of the fundamental frequency of the speech.