DATA COMPRESSION SYSTEM USING FREQUENCY BAND TRANSLATION AND INTERMEDIATE SAMPLE EXTRAPOLATION

Patent №

US 4,852,129

Granted

1989-07-25

Filed 1986

Owner

NIRAVOICE, INC.

Lab

AI components

1

speech

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

06888453

A digital data transmission system for transmitting selected periodic ones of real data samples and intermediate extrapolated data samples. The transmitting end of the communication system includes means for sampling periodic ones of an information signal, an extrapolator for estimating intermediate samples between the selected periodic real samples, correction means for generating a correction value based upon the difference between extrapolated data and real data samples, and transmitting means that transmit periodic ones of real data samples and intermediate correction values. The receiving end includes means for receiving selected real digital data samples and the correction values, and a second extrapolator for estimating intermediate samples between the transmitted real data samples. The receiving end also includes correction means for reconstructing representations of the intermediate data samples computed at the transmitting end based upon the transmitted periodic real data sample and the transmitted correction value. Accordingly, the invention attains bit savings in the communication system. At the both the transmitting and receiving ends, intermediate extrapolated data samples are formed on plural frequency-translated subbands of the lower portion of the voice spectrum. Each low-frequency subband is used for producing the extrapolated value of the respective low-frequency subband. A composite extrapolated signal is formed by combining the results from each subband thereby to reconstruct the intermediate extrapolated sample. Tracking errors between the transmitting extrapolator and the receiving extrapolator are aligned by use of the correction values.

SpeechH04B 14/046H04B 1/66H04J 3/1688

AI classification

Speech0.99
Vision0.01
Planning0.00
AI hardware0.00
Natural language0.00
Machine learning0.00
Evolutionary computation0.00
Knowledge representation0.00

Ownership

NIRAVOICE, INC.

assignment · 50170154

Assignors

SCHWARTZ, NIRA

On an employer assignment, the assignors are typically the inventors.

© 2026 NYSGPT2525 LLC