Speech encoding using a pre-encoded database

Patent metadata

Field	Value
Publication number	US-2021104250-A1
Application number	US-201916591478-A
Country	US
Kind code	A1
Filing date	Oct 2, 2019
Priority date	Oct 2, 2019
Publication date	Apr 8, 2021
Grant date	—

Abstract

Methods, systems, and devices for encoding are described. A device, which may be otherwise known as user equipment (UE), may support standards-compatible audio encoding (e.g., speech encoding) using a pre-encoded database. The device may receive a digital representation of an audio signal and identify, based on receiving the digital representation of the audio signal, a database that is pre-encoded according to a coding standard and that includes a quantity of digital representations of other audio signals. The device may encode the digital representation of the audio signal using a machine learning scheme and information from the database pre-encoded according to the coding standard. The device may generate a bitstream of the digital representation that is compatible with the coding standard based on encoding the digital representation of the audio signal, and output a representation of the bitstream.

First claim

What is claimed is:

1 . A method comprising:
receiving a digital representation of an audio signal; identifying, based at least in part on receiving the digital representation of the audio signal, a database that is pre-encoded according to a coding standard and that comprises a quantity of digital representations of other audio signals; encoding the digital representation of the audio signal using a machine learning scheme and information from the database pre-encoded according to the coding standard; generating a bitstream of the digital representation that is compatible with the coding standard based at least in part on encoding the digital representation of the audio signal; and outputting a representation of the bitstream.

2 . The method of claim 1 , further comprising:
pre-encoding the database according to the coding standard prior to receiving the digital representation of the audio signal; and selecting the pre-encoded database based at least in part on a criterion, wherein identifying the database pre-encoded according to the coding standard is based at least in part on the selecting.

3 . The method of claim 2 , wherein the criterion comprises one or more of a format of the audio signal, a transmission rate associated with a transmission of the audio signal, or a network associated with the transmission of the audio signal.

4 . The method of claim 2 , wherein pre-encoding the database according to the coding standard comprises:
encoding a set of packets according to the coding standard, wherein one or more packets of the set of packets correspond to a database frame in the database; and inserting a set of reset frames between one or more packets of the encoded set of packets.

5 . The method of claim 4 , further comprising:
determining a set of reference points associated with the database based at least in part on the set of packets; and assigning the set of reference points in the database based at least in part on a parameter comprising a distance between reset frames of the set of reset frames, wherein inserting the set of reset frames is based at least in part on the assigning.

6 . The method of claim 5 , further comprising:
selecting a value of the distance from a range of distance values, wherein assigning the set of reference points in the database based at least in part on the selecting.

7 . The method of claim 4 , wherein encoding the digital representation of the audio signal comprises:
ignoring, based at least in part on the set of reset frames, one or more dependencies of a packet of the encoded set of packets with respect to one or more other packets of the encoded set of packets; and encoding a current input frame of the audio signal based at least in part on the ignoring.

8 . The method of claim 4 , further comprising:
determining a set of continuous packets of the encoded set of packets, wherein inserting the set of reset frames between the one or more packets of the encoded set of packets comprises: inserting a first reset frame prior to a first packet of the set of continuous packets of the encoded set of packets; and inserting a second reset frame after a last packet of the set of continuous packets of the encoded set of packets.

9 . The method of claim 2 , further comprising:
determining one or more of a coding mode or a pitch gain associated with the coding standard, wherein pre-encoding the database is based at least in part on one or more of the coding mode or the pitch gain associated with the coding standard.

10 . The method of claim 1 , further comprising:
estimating a scan result associated with the digital representation of the audio signal and the database, wherein encoding the digital representation of the audio signal is based at least in part on the scan result.

11 . The method of claim 10 , further comprising:
training the machine learning scheme to match one or more scanning approach decisions for one or more digital representations of one or more audio signals with respect to the database, wherein estimating the scan result is based at least in part on the training.

12 . The method of claim 1 , wherein encoding the digital representation of the audio signal comprises:
encoding the digital representation jointly according to the coding standard and an additional coding standard different from the coding standard.

13 . The method of claim 1 , further comprising:
receiving a digital representation of a second audio signal; identifying, based at least in part on the receiving of the digital representation of the second audio signal, a set of weighting coefficients of the machine learning scheme, wherein the set of weighting coefficients are associated with an additional coding standard different from the coding standard; encoding the digital representation of the second audio signal using the machine learning scheme based at least in part on one or more weighting coefficients of the set of weighting coefficients; generating a second bitstream of the digital representation of the second audio signal that is compatible with the additional coding standard based at least in part on the encoding of the digital representation of the second audio signal; and outputting a representation of the second bitstream.

14 . An apparatus comprising:
a processor, memory coupled with the processor; and instructions stored in the memory and executable by the processor to cause the apparatus to:
receive a digital representation of an audio signal;
identify, based at least in part on receiving the digital representation of the audio signal, a database that is pre-encoded according to a coding standard and that comprises a quantity of digital representations of other audio signals;
encode the digital representation of the audio signal using a machine learning scheme and information from the database pre-encoded according to the coding standard;
generate a bitstream of the digital representation that is compatible with the coding standard based at least in part on encoding the digital representation of the audio signal; and
output a representation of the bitstream.

15 . The apparatus of claim 14 , wherein the instructions are further executable by the processor to cause the apparatus to:
pre-encode the database according to the coding standard prior to receiving the digital representation of the audio signal; and select the pre-encoded database based at least in part on a criterion, wherein identifying the database pre-encoded according to the coding standard is based at least in part on the selecting.

16 . The apparatus of claim 15 , wherein the instructions to pre-encode the database according to the coding standard are executable by the processor to cause the apparatus to:
encode a set of packets according to the coding standard, wherein one or more packets of the set of packets correspond to a database frame in the database; and insert a set of reset frames between one or more packets of the encoded set of packets.

17 . The apparatus of claim 16 , wherein the instructions are further executable by the processor to cause the apparatus to:
determine a set of reference points associated with the database based at least in part on the set of packets; and assign the set of reference points in the database based at least in part on a parameter comprising a distance between reset frames of the set of reset frames, wherein inserting the set of reset frames is based at least in part on the assigning.

18 . The apparatus of claim 16 , wherein th

Assignees

• Qualcomm Inc

Inventors

• Villette Stephane Pierre
• Sinder Daniel Jared

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G10L2019/0014

  Selection criteria for distances · CPC title

• G10L2019/0013

  Codebook search algorithms · CPC title

• G06N3/08

  Learning methods · CPC title