System and method for modulation domain-based audio signal encoding

1 . A computer-implemented method comprising:

receiving, by an audio encoder, a voice audio signal carrying Personally Identifiable Information (PII) of a speaker;

encoding the voice audio signal to obtain an encoded voice audio signal for securely communicating to a distributed speech processing machine learning (ML) model via a telecommunications network, wherein encoding the voice audio signal includes converting the voice audio signal into the modulation domain to obtain a modulation domain representation of the voice audio signal, encoding the modulation domain representation of the voice audio signal with a plurality of carrier signals and a plurality of modulator signals derived from the modulation domain representation of the voice audio signal, and converting the encoded modulation domain representation of the voice audio signal to the time domain to obtain the encoded voice audio signal; and

transmitting, via the telecommunications network, the encoded voice audio signal to the distributed speech processing ML model.

2 . The computer-implemented method of claim 1 , wherein the distributed speech processing ML model is trained for speech recognition using training data that includes voice audio encoded in the modulation domain.

3 . The computer-implemented method of claim 1 , further comprising:

processing the encoded voice audio signal directly using the distributed speech processing ML model upon reception over the telecommunications network.

4 . The computer-implemented method of claim 3 , wherein encoding the modulation domain representation of the voice audio signal includes processing an encoding key defining an encoding process for the modulation domain representation of the voice audio signal.

5 . The computer-implemented method of claim 4 , wherein decoding the modulation domain representation of the encoded voice audio signal includes processing the encoding key to decode the encoded voice audio signal.

6 . The computer-implemented method of claim 1 , wherein encoding the modulation domain representation of the voice audio signal includes switching the plurality of modulator signals within a plurality of carrier-modulator signal pairs.

7 . The computer-implemented method of claim 6 , wherein switching the plurality of modulator signals within the plurality of carrier-modulator signal pairs includes switching frequency-adjacent modulator signals between the plurality of carrier-modulator signal pairs.

8 . The computer-implemented method of claim 6 , wherein encoding the modulation domain representation of the voice audio signal includes switching the plurality of modulator signals between the plurality of carrier-modulator signal pairs based upon, at least in part, pitch information associated with the voice audio signal.

9 . A computing system comprising:

a processor; and

a memory storing programming instructions for execution by the processor, the programming instructions, upon execution by the processor, causing the computing system to perform the following operations:

receiving, by an audio encoder, a voice audio signal carrying Personally Identifiable Information (PII) of a speaker;

encoding the voice audio signal to obtain an encoded voice audio signal for securely communicating to a distributed speech processing machine learning (ML) model via a telecommunications network, wherein encoding the voice audio signal includes converting the voice audio signal into the modulation domain to obtain a modulation domain representation of the voice audio signal, encoding the modulation domain representation of the voice audio signal with a plurality of carrier signals and a plurality of modulator signals derived from the modulation domain representation of the voice audio signal, and converting the encoded modulation domain representation of the voice audio signal to the time domain to obtain the encoded voice audio signal; and

transmitting, via the telecommunications network, the encoded voice audio signal to the distributed speech processing ML model.

10 . The computing system of claim 9 , wherein the distributed speech processing ML model is trained for speech recognition using training data that includes voice audio encoded in the modulation domain.

11 . The computing system of claim 9 , wherein encoding the modulation domain representation of the voice audio signal includes switching the plurality of modulator signals within a plurality of carrier-modulator signal pairs.

12 . The computing system of claim 11 , wherein switching the plurality of modulator signals within the plurality of carrier-modulator signal pairs includes switching frequency-adjacent modulator signals between the plurality of carrier-modulator signal pairs.

13 . The computing system of claim 12 , wherein encoding the modulation domain representation of the voice audio signal includes switching the plurality of modulator signals between the plurality of carrier-modulator signal pairs based upon, at least in part, pitch information associated with the voice audio signal.

14 . A computer program product residing on a non-transitory computer readable medium having programming instructions stored thereon which, when executed by a processor of a system, cause the system to perform the following operations:

receiving, by an audio encoder, a voice audio signal carrying Personally Identifiable Information (PII) of a speaker;

encoding the voice audio signal to obtain an encoded voice audio signal for securely communicating to a distributed speech processing machine learning (ML) model via a telecommunications network, wherein encoding the voice audio signal includes converting the voice audio signal into the modulation domain to obtain a modulation domain representation of the voice audio signal, encoding the modulation domain representation of the voice audio signal with a plurality of carrier signals and a plurality of modulator signals derived from the modulation domain representation of the voice audio signal to obtain the encoded voice audio signal, and converting the encoded modulation domain representation of the voice audio signal to the time domain to obtain the encoded voice audio signal; and

transmitting, via the telecommunications network, the encoded voice audio signal to the distributed speech processing ML model.

15 . The computer program product of claim 14 , wherein the distributed speech processing ML model is trained for speech recognition using training data that includes voice audio encoded in the modulation domain.

16 . The computer program product of claim 14 , wherein encoding the modulation domain representation of the voice audio signal includes processing an encoding key defining an encoding process for the modulation domain representation of the voice audio signal.

17 . The computer program product of claim 16 , wherein decoding the modulation domain representation of the voice audio signal includes processing the encoding key to decode the modulation domain representation of the voice audio signal.

18 . The computer program product of claim 14 , wherein decoding the modulation domain representation of the voice audio signal includes switching the plurality of modulator signals within a plurality of carrier-modulator signal pairs.

19 . The computer program product of claim 18 , wherein switching the plurality of modulator signals within the plurality of carrier-modulator signal pairs includes switching frequency-adjacent modulator signals between the plurality of carrier-modulator signal pairs.

20 . The computer program product of claim 18 , wherein encoding the modulation domain representation of the voice audio signal includes switching the plurality of modulator signals between the plurality of carrier-modulator signal pairs based upon, at least in part, pitch information associated with the voice audio signal.