Method for improving far-field speech interaction performance, and far-field speech interaction system

A non-transitory computer-readable storage medium with a computer program stored thereon, wherein when executed by a processor, the computer program causes the processor to cyclically perform steps S1 to S6 by N times, wherein N is a natural number not less than 2:

• step S1:
• receiving, by a microphone of a smart device, a network configuration soundwave signal of a predetermined length of time during a network configuration process of the smart device;
• step S2:
• performing front-end speech processing on the network configuration soundwave signal based on front-end speech processing parameter, wherein the front-end speech processing comprises echo cancellation, and the front-end speech processing parameter comprises a length of an echo cancellation filter;
• step S3:
• calculating an energy value En of the network configuration soundwave signal processed by the front-end speech processing in step S2 within the current cycle;
• step S4:
• determining whether the current cycle is a first cycle, and in response to that the current cycle is the first cycle, calculating an initial spatial acoustic parameter Pinitial based on the network configuration soundwave signal received in step S1, and proceeding to step S5;
• and in response to that the current cycle is not the first cycle, proceeding directly to step S5;
• step S5:
• determining a numerical value of a spatial acoustic parameter Pn+1 in a next cycle by combining a preset upper limit margin and a preset lower limit margin of the smart device, such that the numerical value of Pn+1 satisfies (Pinitial−M1)≤Pn+1≤(Pinitial+M2), wherein n is a sequence number of the current cycle, M1 is the preset lower limit margin, and M2 is the preset upper limit margin;
• step S6:
• updating the front-end speech processing parameter using the spatial acoustic parameter Pn+1 obtained in step S5 for the next cycle;
• after that the N cycles are completed, proceeding to step S7:
• determining a minimum value Emin among the network configuration soundwave signal energy values obtained in the N cycles, and taking a spatial acoustic parameter in the cycle corresponding to the minimum value Emin as a final spatial acoustic parameter;
• wherein the final spatial acoustic parameter is configured to update the length of the echo cancellation filter to improve far-field speech interaction performance.

A terminal device comprising:

• a microphone;
• a processor;
• and
• a memory;
• wherein the memory is configured to store a computer program, which, when executed by the processor, causes the processor to cyclically perform steps S1 to S6 by N times, wherein N is a natural number not less than 2:
• step S1:
• receiving, by the microphone, a network configuration soundwave signal of a predetermined length of time during a network configuration process of the terminal device;
• step S2:
• performing front-end speech processing on the network configuration soundwave signal based on front-end speech processing parameter;
• wherein the front-end speech processing comprises echo cancellation, and the front-end speech processing parameter comprises a length of an echo cancellation filter;
• step S3:
• calculating an energy value En of the network configuration soundwave signal processed by the front-end speech processing in step S2 within the current cycle;
• step S4:
• determining whether the current cycle is a first cycle, and in response to that the current cycle is the first cycle, calculating an initial spatial acoustic parameter Pinitial based on the network configuration soundwave signal received in step S1, and proceeding to step S5;
• and in response to that the current cycle is not the first cycle, proceeding directly to step S5;
• step S5:
• determining a numerical value of a spatial acoustic parameter Pn+1 in a next cycle by combining a preset upper limit margin and a preset lower limit margin of the terminal device, such that the numerical value of Pn+1 satisfies (Pinitial−M1)≤Pn+1≤(Pinitial+M2), wherein n is a sequence number of the current cycle, M1 is the preset lower limit margin, and M2 is the preset upper limit margin;
• step S6:
• updating the front-end speech processing parameter using the spatial acoustic parameter Pn+1 obtained in step S5 for the next cycle;
• after that the N cycles are completed, proceeding to step S7:
• determining a minimum value Emin among the network configuration soundwave signal energy values obtained in the N cycles, and taking a spatial acoustic parameter in the cycle corresponding to the minimum value Emin as a final spatial acoustic parameter;
• wherein the final spatial acoustic parameter is configured to update the length of the echo cancellation filter to improve far-field speech interaction performance.