METHOD AND SYSTEM FOR SPEECH SYNTHESIS

• US 20090299746A1
• Filed: 05/27/2009
• Published: 12/03/2009
• Est. Priority Date: 05/28/2008
• Status: Active Grant

• Current Assignee: Cerence Operating Company (Cerence Inc.)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Qin, Yong, Shuang, Zhiwei, Shi, Qin, Meng, Fan Ping
• Granted Patent: US 8,321,223 B2
• Time in Patent Office: Days
• US Class Current: 704/260
• CPC Class Codes: G10L 13/04 Details of speech synthesis systems, e.g. synthesiser structure or memory management
• 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 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

A method for performing speech synthesis to a textual content at a client. The method includes the steps of: performing speech synthesis to the textual content based on a current acoustical unit set S_current in a corpus at the client; analyzing the textual content and generating a list of target units with corresponding context features, selecting multiple acoustical unit candidates for each target unit according to the context features based on an acoustical unit set S_total that is more plentiful than the current acoustical unit set S_current in the corpus at the client, and determining acoustical units suitable for speech synthesis for the textual content according to the multiple unit candidates; and updating the current acoustical unit set S_current in the corpus at the client based on the determined acoustical units.

Claims

1. A method for performing speech synthesis to a textual content at a client data processing system, the method comprising:

performing speech synthesis to the textual content based on a current acoustical unit set S_current in a corpus at said client;

analyzing said textual content and generating a list of target units with corresponding context features;

selecting multiple acoustical unit candidates for each target unit according to said context features based on an acoustical unit set S_total that is more plentiful than the current acoustical unit set S_current in the corpus at said client;

determining acoustical units suitable for speech synthesis for said textual content according to said multiple unit candidates; and

updating the current acoustical unit set S_current in the corpus at said client based on the determined acoustical units.

2. The method according to claim 1, further comprising the step of:

downloading a set S₀ of a small number of acoustical units, which can perform speech synthesis to all kinds of textual contents and which can ensure an acceptable speech synthesis quality, as an initial current acoustical unit set in the corpus on said client to make S_current=S₀.

3. The method according to claim 1, wherein said step of determining acoustical units further comprises:

ranking said multiple acoustical unit candidates to determine, according to importance for the textual content, an acoustical unit set for updating the current acoustical unit set in the corpus at said client.

4. The method according to claim 3, further comprising the step of:

downloading into said client an acoustical unit set S_Δ

which (i) belongs to the acoustical unit set used for update and (ii) is not included in the current acoustical unit set in the corpus at said client; and

wherein the current acoustical unit set S_current in the corpus on said client is updated by making S_current=S_current+S_Δ

in said updating step.

5. The method according to claim 3, wherein the unit candidates are ranked based on how many times each unit candidate has been selected.

6. The method according to claim 5, wherein multiple acoustical unit candidates of different target units are ranked together.

7. The method according to claim 5, wherein multiple acoustical unit candidates of each target unit are ranked separately.

8. A system for enabling speech synthesis to a textual content at a client data processing system, the system comprising:

speech synthesis means configured to perform speech synthesis to the textual content based on a current acoustical unit set S_current in a corpus on said client;

analysis means configured to analyze said textual content and generate a list of target units with corresponding context features;

selection means configured to select multiple acoustical unit candidates for each target unit according to said context features based on an acoustical unit set S_total that is more plentiful than the current acoustical unit set S_current in the corpus at said client;

determining means configured to determine acoustical units suitable for speech synthesis for said textual content according to said multiple unit candidates; and

update means configured to update the current acoustical unit set S_current in the corpus on said client based at the determined acoustical units.

9. The system according to claim 8, further comprising:

means configured to download a set S₀ of a small number of acoustical units which can perform speech synthesis to all kinds of textual contents and which can ensure an acceptable speech synthesis quality, as an initial current acoustical unit set in the corpus on said client to make S_current=S₀.

10. The system according to claim 8, wherein said determining means comprises:

means to rank said multiple acoustical unit candidates to determine, according to importance for the textual content, an acoustical unit set for updating the current acoustical unit set in the corpus at said client.

11. The system according to claim 10, wherein said determining means further comprises:

means for determining an acoustical unit set S_Δ

which (i) belongs to the acoustical unit set used for update and (ii) is not included in the current acoustical unit set in the corpus at said client; and

wherein said update means is configured to update the current acoustical unit set S_current in the corpus on said client by making S_current =S_current+S_Δ

.

12. The system according to claim 10, wherein said determining means is configured to rank the unit candidates based on how many times each unit candidate has been selected.

13. The system according to claim 12, wherein said determining means comprises means to rank multiple acoustical unit candidates of different target units together.

14. The system according to claim 12, wherein said determining means comprises means to separately rank multiple acoustical unit candidates of each target unit.