Prosodic number string synthesis

• US 5,740,319 A
• Filed: 11/24/1993
• Issued: 04/14/1998
• Est. Priority Date: 11/24/1993
• Status: Expired due to Term

• Current Assignee: Intel Corporation
• Original Assignee: Texas Instruments, Inc.
• Inventors: Wedemeier, Frederick C.
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): Dorvil, Richemond
• Application Number:
  US08/157,791
• Time in Patent Office: 1,602 Days
• Field of Search: 395/2.79, 395/2.81, 395/2.67, 395/2.76, 395/2.69, 395/2.7, 395/2.75, 381/39
• US Class Current: 704/266
• CPC Class Codes: G10L 13/06 Elementary speech units used in speech synthesisers; Concatenation rules

Abstract

A machine and method for providing human voice sounding numbers includes storing in digital form segments of leading digit utterance, segments of trailing digit utterances, group pausing utterances and digit pair utterances. A data string of segments is read out of storage and concatenated.

Claims

1. A synthesizer for a human voice for numbers comprising:

means for storing human voiced leading-digit utterance segments, human voiced trailing-digit utterance segments, human voiced digit group pause segments, and human voiced digit pair utterance segments; and

means coupled to said storage means and responsive to a data string of numbers for reading out for each digit a pair of said stored human voiced segments according to said string of numbers to produce a natural sound of a human voice.

2. A synthesizer for human voice for numbers comprising:

storage means for storing in digital form human voiced leading-digit utterance segments, human voiced trailing-digit utterance segments, human voiced group pausing utterance segments and human voiced digit-pair utterance segments;

means coupled to said storage means and responsive to a data string of numbers for reading out and concatenating said human voiced segments according to said data string of numbers; and

digital-to-analog generator means responsive to said segments for providing natural sounding human voice speech output.

3. A method of providing a human voice sounding string of numbers comprising:

storing human voiced leading-digit utterance segments, trailing-digit utterance segments, group-pausing utterance segments, and digit-pair digit utterance segments; and

reading out said stored segments of human voiced digit utterances according to a data string of numbers to produce natural sounding human voice speech.

4. The method of claim 3 wherein said storing step includes the steps of:

storing human voiced leading-digit utterances with constant pitch, cadence and volume;

storing human voiced trailing-digit utterances with constant pitch, cadence and volume;

storing human voiced group-pausing utterances with constant pitch, cadence and volume; and

storing human voiced digit-pair utterances with constant pitch, cadence and volume.

5. A method of providing synthesized voice output of a numeric string comprising the steps of:

recording selected samples of actual human voice spoken numbers;

segmenting said actual human voiced spoken numbers into subdigit speech segments of more than one human voiced of leading-digit utterances, human voiced trailing-digit utterances, human voiced group-pausing or human voiced digit-pair utterances; and

combining at least two of said subdigit speech segments according to desired spoken numeric string output to produce a natural sound of a human voice.

6. A method of providing a synthesized voice output of any numeric string comprising the steps of:

recording selected samples of actual human voiced spoken numbers,

segmenting said actual human voiced spoken numbers into 130 subdigit speech segments including all digits 0 through 9 at the beginning of a string, all digits 0 through 9 at the end of a string, all digits 0 through 9 indicating a spoken pause, and all digit pairs from 00 to 99; and

combining said subdigit speech segments to produce a natural sound of a human voice.