Synthetic narrowband data generation for narrowband automatic speech recognition systems
A system and apparatus are provided for generating synthetic telephony narrowband data for training an automatic speech recognition model by receiving a broadband audio data file and then initiating a telephony call using a pre-configured telephone provider to play the broadband audio data file in the telephony call and to record and store audio data generated by transmission of the broadband audio data file in the telephony call, thereby generating the synthetic telephony narrowband data file from the broadband audio data file.
1. An information handling system comprising:
one or more processors;
a memory coupled to at least one of the processors;
a set of instructions stored in the memory and executed by at least one of the processors to generate synthetic telephony narrowband data from a broadband audio data file, wherein the set of instructions are executable to perform actions of:
monitoring, by the system, a first repository to detect a new broadband audio data file;
upon detecting the new broadband audio data file, initiating, by the system, a telephony call that is placed using a pre-configured telephone provider to a designated phone number that can be received and/or monitored by the system;
playing, by the system, the new broadband audio data file for transmission in the telephony call;
receiving, by the system, the telephony call at the designated phone number to capture audio data generated by transmitting the new broadband audio data file in the telephony call;
recording, by the system, the audio data as a synthetic telephony narrowband data file corresponding to the new broadband audio data file project;
storing, by the system, the synthetic telephony narrowband data file as an entry in a second repository; and
terminating, by the system, the telephony call placed by the pre-configured telephone provider.
2. The information handling system of claim 1 , wherein the set of instructions are executable to train an automatic speech recognition model using both the new broadband audio data file and the synthetic telephony narrowband data file.
3. The information handling system of claim 1 , where the second repository is the first repository.
4. The information handling system of claim 1 , wherein the set of instructions are executable to initiate the telephony call by programmatically placing the telephony call through a telephone provider for playback and recording of the new broadband audio data file by the system.
5. The information handling system of claim 1 , where the pre-configured telephone provider comprises a telecommunications provider, a cloud communications platform, a voice over IP service, or a telephony infrastructure platform that provides connectivity between HTTP and a public switched telephone network (PSTN).
6. The information handling system of claim 1 , wherein the set of instructions are executable to transmit the new broadband audio data file in the telephony call to directly compress all of the new broadband audio data file using the telephony call to generate a synthetic telephony narrowband data file.
7. A computer program product stored in a computer readable storage medium, comprising computer instructions that, when executed by a processor at an information handling system, causes the system to generate synthetic telephony narrowband data from a broadband audio data file for training an automatic speech recognition model by:
monitoring, by the system, a first repository to detect a new broadband audio data file;
upon detecting the new broadband audio data file, initiating, by the system, a telephony call that is placed using a pre-configured telephone provider to a designated phone number that can be received and/or monitored by the system;
receiving, by the system, the telephony call at the designated phone number to capture audio data generated by transmitting the new broadband audio data file in the telephony call;
recording, by the system, the audio data as a synthetic telephony narrowband data file corresponding to the new broadband audio data file project;
storing, by the system, the synthetic telephony narrowband data file as an entry in a second repository; and
terminating, by the system, the telephony call placed by the pre-configured telephone provider.
8. The computer program product of claim 7 , further comprising computer instructions that, when executed by the system, cause the system to train the automatic speech recognition model using both the new broadband audio data file and the synthetic telephony narrowband data file.
9. The computer program product of claim 7 , further comprising computer instructions that, when executed by the system, cause the system to initiate the telephony call by programmatically placing the telephony call through a telephone provider for playback and recording of the new broadband audio data file by the system.
10. The computer program product of claim 7 , where the computer instructions are executed by the system to initiate the telephony call by placing the telephony call using a telecommunications provider, a cloud communications platform, a voice over IP service, or a telephony infrastructure platform that provides connectivity between HTTP and a public switched telephone network (PSTN).
11. The computer program product of claim 7 , where the computer instructions are executed by the system to transmit the new broadband audio data file in the telephony call to directly compress all of the new broadband audio data file using the telephony call to generate a synthetic telephony narrowband data file.