Time compression/expansion of phonemes based on the information carrying elements of the phonemes

• US 5,729,657 A
• Filed: 04/16/1997
• Issued: 03/17/1998
• Est. Priority Date: 11/25/1993
• Status: Expired due to Term

• Current Assignee: Intellectual Ventures I LLC (Intellectual Ventures LLC)
• Original Assignee: Telia AB (Government of Norway)
• Inventors: Svensson, Tomas
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): MATTSON, ROBERT
• Application Number:
  US08/834,391
• Time in Patent Office: 335 Days
• Field of Search: 395/2.67, 395/2.69, 395/2.73, 395/2.76
• US Class Current: 704/267
• CPC Class Codes: G10L 13/06 Elementary speech units used in speech synthesisers; Concatenation rules
• CPC Class Codes: G10L 13/08 Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination
• CPC Class Codes: G10L 21/04 Time compression or expansion

Abstract

The present invention relates to a method and arrangement for transforming phonemes over a shorter or longer time than an existing phoneme. The transformation takes place asymmetrically in that a basic phoneme is divided into a number of points, the said points being identified with respect to information-carrying elements in the phoneme. This provides a weighting in the phoneme between information-carrying elements and elements carrying less information. The parts of the phoneme which elements carrying less information are transformed over a longer or, respectively, shorter time interval. Elements in the phoneme which represent information-carrying parts are transferred unchanged in time. This provides a transformation of the phoneme which retains its original character in all essentials.

By the parts of the phoneme carrying less information being identified, the invention also provides an indication of where different phonemes can be fitted into one another in the creation of artificial speech.

Claims

1. A speech-synthesis method for transforming a phoneme from a first timescale to a second timescale, comprising the steps of:

determining a set of points indicative of said phoneme;

identifying a first part of said set of points occurring at a boundary of said phoneme and having a first amount of information uniquely characterizing said phoneme, said first part corresponding to a first period in said first timescale;

identifying a second part of said set of points occurring at an interior of said phoneme and having a lesser amount of information than said first part, said second part corresponding to a second period in said first timescale;

transforming said second part to said second timescale to create a transformed second part having a third period that is different than said second period; and

transforming said first part to said second timescale to create a transformed first part having a fourth period that is equivalent or nearly equivalent to said first period so as to retain said information carried by said first part not carried by said second part.

2. The method of claim 1, further comprising the steps of:

determining an amount of said information carried by respective of said set of points; and

weighting respective of said set of points based on said amount of information carried by said respective of said set of points.

3. The method of claim 2, wherein:

said step of weighting comprises weighting said second part with respective lower weighting values than said first part; and

said step of transforming said second part comprises,

duplicating a first portion of said second part of said set of points when said second timescale is longer in duration than said first timescale, and

removing a second portion of said second part of said set of points when said second timescale is shorter in duration that said first timescale.

4. The method of claim 1, further comprising:

combining said phoneme with another phoneme, comprising,

identifying a part of said another phoneme which carries nearly no information, and

transitioning from said phoneme to said another phoneme at said part of said another phoneme which carries nearly no information.

5. The method of claim 1, wherein:

said step of determining a set of points indicative of said phoneme comprises determining a fundamental tone of said phoneme; and

said step of transforming said second part comprises transforming said second part to said second timescale only when a duration of said first timescale does not equal a duration of said second timescale thereby retaining said fundamental tone.

6. A speech synthesis system that transforms a phoneme from a first timescale to a second timescale, comprising:

a selection element configured to select said phoneme from at least one of a speech sequence or a storage device;

a determination mechanism configured to determine a set of points indicative of said phoneme, said determination mechanism comprising,

a first identification mechanism that identifies a first part of said set of points occurring at a boundary of said phoneme and having a first amount of information uniquely characterizing said phoneme and corresponding to a first period in said first timescale, and

a second identification mechanism that identifies a second part of said set of points occurring at an interior of said phoneme having a lesser amount of information than said second part and corresponding to a second period in said first timescale;

a first transforming mechanism that transforms said second part to said second timescale to create a transformed second part having a third period that is different than said second period; and

a second transforming mechanism that transforms said first part to said second timescale to create a transformed first part having a fourth period equivalent or nearly equivalent to said first period so as to retain information carried by said first part not carried by said second part.

7. The system of claim 6, wherein said selection element is configured to determine an amount of said information carried by respective of said set of points, and is configured to weight respective of said set of points based on said amount of information carried by respective of said set of points.

8. The system of claim 7, wherein:

said selection element weights said second part with respective lower weighting values than said first part, and weights with medium weights a third part of said set of points carrying more information than said second part but less than said first part; and

said first transforming mechanism is configured to transform said second part over a longer timescale than said third part, and transform said third part over a longer timescale than said first part.

9. The system of claim 7, wherein:

said selection element weights said second part with respective lower weighting values than said first part, and weights with medium weights a third part of said set of points carrying more information than said second part but less than said first part; and

said first transforming mechanism is configured to transform three points of said second part into a corresponding single point, and transform two points of said third part over a timescale that is shorter in duration than said first part.

10. The system of claim 6, wherein:

said selection element determines a fundamental tone of said phoneme based on said set of points; and

said first transforming mechanism is configured to change said fundamental tone only when said second timescale is different than said first timescale.