Method, apparatus and computer program product for providing text independent voice conversion

Abstract

An apparatus for providing text independent voice conversion may include a first voice conversion model and a second voice conversion model. The first voice conversion model may be trained with respect to conversion of training source speech to synthetic speech corresponding to the training source speech. The second voice conversion model may be trained with respect to conversion to training target speech from synthetic speech corresponding to the training target speech. An output of the first voice conversion model may be communicated to the second voice conversion model to process source speech input into the first voice conversion model into target speech corresponding to the source speech as the output of the second voice conversion model.

Description

CROSS REFERENCE TO RELATED APPLICATION[0001]This patent application is a continuation application of US Publication No. 2009 / 0094031, published on Apr. 9, 2009 (U.S. patent application Ser. No. 11 / 867,196 filed on Oct. 4, 2007), the subject matter of which is herein incorporated by reference.FIELD OF THE INVENTION[0002]Embodiments of the present invention relate generally to voice conversion technology and, more particularly, relate to a method, apparatus and computer program product for providing voice conversion that is text independent.BACKGROUND OF THE INVENTION[0003]The modern communications era has brought about a tremendous expansion of wireline and wireless networks. Computer networks, television networks, and telephony networks are experiencing an unprecedented technological expansion, fueled by consumer demand. Wireless and mobile networking technologies have addressed related consumer demands, while providing more flexibility and immediacy of information transfer.[0004]Cu...

AI Technical Summary

Benefits of technology

The present invention offers a way to convert audio to high-quality voice in a way that is easier to use than existing methods. This allows for simultaneous conversion of multiple source speakers to target speakers using the same conversion mechanism.

Problems solved by technology

However, depending upon the target speaker's identity and other factors, it may sometimes be difficult or impossible to collect a parallel corpus sufficient for voice conversion between a particular pair of source and target speakers.

Furthermore, free speech (i.e., unscripted speech) recorded from either the source or target speaker is often not useful for use in voice conversion, since there is not necessarily a match between the words and / or phrases spoken by the source and target speakers in free speech.

However, to date, voice conversion techniques that do not rely on a parallel corpus for training typically perform worse than parallel corpus schemes.

Furthermore, such schemes typically require linguistic knowledge for system tuning and very large databases in order to find parallel subunits from both source and target speakers within a certain context.

Particularly in mobile environments, increases in memory consumption directly affect the cost of devices employing such methods.

Images