Receiver side noise reduction and speech enhancement for PSTN callers
In one embodiment, an illustrative method herein may comprise: receiving, at a receiver, an audio codec stream; determining, by the receiver, a length of time associated with a look-ahead buffer of the audio codec stream; inputting, by the receiver, the audio codec stream into an audio enhancement model trained with one or more audio enhancements to cause the audio enhancement model to apply the one or more audio enhancements to the audio codec stream to generate an enhanced audio codec stream within the length of time associated with the look-ahead buffer of the audio codec stream; and outputting, by the receiver, the enhanced audio codec stream.
1 . A method, comprising:
receiving, at a receiver, an audio codec stream, wherein the audio codec stream originates from a public switched telephone network (PSTN) caller;
determining, by the receiver and in real-time, a length of time associated with a look-ahead buffer of the audio codec stream;
inputting, by the receiver and in response to the audio codec stream originating from a public switched telephone network (PSTN) caller, the audio codec stream into an audio enhancement model trained with one or more audio enhancements to cause the audio enhancement model to apply the one or more audio enhancements to the audio codec stream to generate an enhanced audio codec stream specifically within the length of time associated with the look-ahead buffer of the audio codec stream as determined in real-time; and
outputting, by the receiver, the enhanced audio codec stream.
2 . The method as in claim 1 , wherein the one or more audio enhancements are selected from a group consisting of: noise reduction; bandwidth extension; artifact rectification;
and packet loss concealment.
3 . The method as in claim 1 , further comprising:
determining a particular user hearing profile; and
inputting the particular user hearing profile into the audio enhancement model to generate the enhanced audio codec stream specifically for the particular user hearing profile.
4 . The method as in claim 1 , further comprising:
determining ambient receiver sound;
determining audio impairments caused by the ambient receiver sound; and
counteracting the audio impairments with the audio enhancement model to generate the enhanced audio codec stream specifically based on the audio impairments.
5 . The method as in claim 1 , further comprising:
determining the one or more audio enhancements from a plurality of optional audio enhancements; and
selecting a particular model trained specifically on the one or more audio enhancements as the audio enhancement model.
6 . The method as in claim 1 , wherein the audio enhancement model is trained based on a plurality of optional audio enhancements, the method further comprising:
entering a selection of the one or more audio enhancements into the audio enhancement model to cause the audio enhancement model to specifically apply the one or more audio enhancements from the plurality of optional audio enhancements.
7 . The method as in claim 1 , wherein the length of time associated with the look-ahead buffer is based on a variable jitter buffer, the method further comprising:
calculating the length of time based on one or more real-time factors of the audio codec stream.
8 . The method as in claim 7 , wherein the one or more real-time factors are selected from a group consisting of: an amount of noise in the audio codec stream; an amount of packet loss in the audio codec stream; a number of audio enhancements to be applied to the audio codec stream; an amount of latency in the audio codec stream; and a processing capacity of the receiver during the audio codec stream.
9 . The method as in claim 7 , wherein the audio enhancement model is trained with a variable length of time, and wherein the audio enhancement model processes the audio codec stream based on the length of time associated with the look-ahead buffer according to the variable jitter buffer.
10 . The method as in claim 9 , wherein the audio enhancement model further processes the audio codec stream based on a number and/or type of the one or more audio enhancements to be applied to the audio codec stream within the length of time associated with the look-ahead buffer according to the variable jitter buffer.
11 . The method as in claim 1 , further comprising:
detecting one or more public switched telephone network callers participating in a conference call; and
activating the audio enhancement model to generate the enhanced audio codec stream for the conference call in response to having at least one public switched telephone network caller.
12 . The method as in claim 1 , further comprising:
activating the audio enhancement model to generate the enhanced audio codec stream for only the audio codec stream of a selected presenter on a conference call.
13 . The method as in claim 1 , wherein the receiver is selected from a group consisting of: a conference call server; a conference call gateway; a conference call end user application; a video call server; a video call gateway; a video call end user application; and a one-on-one call end user receiver.
14 . A tangible, non-transitory, computer-readable medium having computer-executable instructions stored thereon that, when executed by a processor on a computer, cause the computer to perform a method comprising:
receiving, at a receiver, an audio codec stream, wherein the audio codec stream originates from a public switched telephone network (PSTN) caller;
determining, by the receiver and in real-time, a length of time associated with a look-ahead buffer of the audio codec stream;
inputting, by the receiver and in response to the audio codec stream originating from a public switched telephone network (PSTN) caller, the audio codec stream into an audio enhancement model trained with one or more audio enhancements to cause the audio enhancement model to apply the one or more audio enhancements to the audio codec stream to generate an enhanced audio codec stream specifically within the length of time associated with the look-ahead buffer of the audio codec stream as determined in real-time; and
outputting, by the receiver, the enhanced audio codec stream.
15 . The tangible, non-transitory, computer-readable medium as in claim 14 , wherein the one or more audio enhancements are selected from a group consisting of: noise reduction;
bandwidth extension; artifact rectification; and packet loss concealment.
16 . The tangible, non-transitory, computer-readable medium as in claim 14 , wherein the method further comprises:
determining a particular user hearing profile; and
inputting the particular user hearing profile into the audio enhancement model to generate the enhanced audio codec stream specifically for the particular user hearing profile.
17 . The tangible, non-transitory, computer-readable medium as in claim 14 , wherein the method further comprises:
determining ambient receiver sound;
determining audio impairments caused by the ambient receiver sound; and
counteracting the audio impairments with the audio enhancement model to generate the enhanced audio codec stream specifically based on the audio impairments.
18 . The tangible, non-transitory, computer-readable medium as in claim 14 , wherein the method further comprises one of either:
determining the one or more audio enhancements from a plurality of optional audio enhancements and selecting a particular model trained specifically on the one or more audio enhancements as the audio enhancement model; or
entering a selection of the one or more audio enhancements into the audio enhancement model to cause the audio enhancement model to specifically apply the one or more audio enhancements from the plurality of optional audio enhancements.
19 . The tangible, non-transitory, computer-readable medium as in claim 14 , wherein the length of time associated with the look-ahead buffer is based on a variable jitter buffer, and wherein the method further comprises:
calculating the length of time based on one or more real-time factors of the audio codec stream.
20 . An apparatus, comprising:
one or more network interfaces to communicate with a network;
a processor coupled to the one or more network interfaces and configured to execute one or more processes; and
a memory configured to store a process that is executable by the processor, the process, when executed, configured to:
receive, as a receiver, an audio codec stream, wherein the audio codec stream originates from a public switched telephone network (PSTN) caller;
determine, as the receiver and in real-time, a length of time associated with a look-ahead buffer of the audio codec stream;
input, as the receiver and in response to the audio codec stream originating from a public switched telephone network (PSTN) caller, the audio codec stream into an audio enhancement model trained with one or more audio enhancements to cause the audio enhancement model to apply the one or more audio enhancements to the audio codec stream to generate an enhanced audio codec stream specifically within the length of time associated with the look-ahead buffer of the audio codec stream as determined in real-time; and
output, as the receiver, the enhanced audio codec stream.