Voice recognition method and apparatus, and air conditioner

1. A voice recognition method, comprising:

acquiring first voice data;

adjusting, according to the first voice data, a collection state of second voice data to obtain an adjusted collection state, and acquiring the second voice data based on the adjusted collection state; and

performing far-field voice recognition on the second voice data using a preset far-field voice recognition model so as to obtain semantic information corresponding to the second voice data,

wherein the adjusting, according to the first voice data, a collection state of second voice data comprises: after determining location information of a sound source sending the first voice data, performing at least one of the following:

enhancing an intensity of the voice collection device that collects the first voice data to collect the second voice data on the location information; and

suppressing an intensity of the voice collection device that collects the first voice data to collect the second voice data at locations except the location information,

wherein the determining location information of a sound source sending the first voice data comprises:

determining a direction of the sound source sending the first voice data using the voice collection device; and

locating the sound source based on the direction by using a locating device to obtain the location information of the sound source,

wherein the locating device comprises: a microwave radar module; and the location information comprises: a distance and a direction.

2. The method of claim 1 , wherein the first voice data comprises a voice wake-up word, the voice wake-up word is voice data for waking up a voice device; and

the second voice data comprises a voice command, the voice command is voice data for controlling the voice device.

3. The method of claim 1 , wherein an operation of acquiring first voice data, an operation of adjusting, according to the first voice data, a collection state of second voice data to obtain an adjusted collection state, and an operation of acquiring the second voice data based on the adjusted collection state are executed at a local side of a voice device; and

an operation of performing far-field voice recognition on the second voice data using the preset far-field voice recognition model is executed by the voice device at the local side in response to reception of feedback information obtained by an operation of cloud.

4. The method of claim 1 , wherein the acquiring first voice data comprises:

acquiring the first voice data collected by a voice collection device;

wherein the acquiring the second voice data comprises:

acquiring the second voice data collected by a voice collection device after the collection state is adjusted,

wherein the voice collection device comprises: a microphone array provided with more than one microphone for collecting voice data in more than one direction.

5. The method of claim 1 , wherein the enhancing an intensity of the voice collection device that collects the first voice data to collect the second voice data on the location information comprises at least one of the following:

in a case that the voice collection device comprises a microphone array, turning on a microphone, on the location information, in the microphone array; and

in a case that the voice collection device comprises the microphone array, increasing the number of microphones turned on, on the location information, in the microphone array.

6. The method of claim 1 , wherein the suppressing an intensity of the voice collection device that collects the first voice data to collect the second voice data at locations except the location information comprises at least one of the following:

turning off microphones at locations, except the location information, in the microphone array; and

reducing the number of the microphones turned on, at locations, except the location information, in the microphone array.

7. The method of claim 1 , wherein the performing far-field voice recognition on the second voice data using the preset far-field voice recognition model comprises:

preprocessing the second voice data to obtain voice information; and

performing, using the preset far-field voice recognition model, far-field voice recognition on the voice information obtained by preprocessing;

wherein the far-field voice recognition model comprises: a far-field acoustic model obtained by deep learning training based on a long short-term memory network (LSTM) algorithm.

8. The method of claim 1 , further comprising:

collecting voice data and sound source data of the voice data; and

performing training, after the voice data and the sound source data is preprocessed, with an LSTM model to obtain a far-field voice recognition model based on LSTM.

9. A voice recognition apparatus, comprising:

a processor for executing a plurality of instructions;

a non-transitory storage medium for storing the plurality of instructions,

wherein the plurality of instructions cause the processor to implement following operations:

acquiring first voice data;

adjusting, according to the first voice data, a collection state of second voice data to obtain an adjusted collection state, and acquire the second voice data based on the adjusted collection state; and

performing far-field voice recognition on the second voice data using a preset far-field voice recognition model so as to obtain semantic information corresponding to the second voice data,

wherein the adjusting, according to the first voice data, a collection state of second voice data comprises: after determining location information of a sound source sending the first voice data, performing at least one of the following:

enhancing an intensity of the voice collection device that collects the first voice data to collect the second voice data on the location information; and

suppressing an intensity of the voice collection device that collects the first voice data to collect the second voice data at locations except the location information,

wherein the determining location information of a sound source sending the first voice data comprises:

determining a direction of the sound source sending the first voice data using the voice collection device; and

locating the sound source based on the direction by using a locating device to obtain the location information of the sound source,

wherein the locating device comprises: a microwave radar module; and the location information comprises: a distance and a direction.

10. The apparatus of claim 9 , wherein the first voice data comprises: a voice wake-up word; the voice wake-up word is voice data for waking up a voice device; and

the second voice data comprises a voice command, the voice command is voice data for controlling the voice device.

11. The apparatus of claim 9 , wherein an operation of acquiring first voice data, an operation of adjusting, according to the first voice data, a collection state of second voice data to obtain an adjusted collection state, and an operation of acquiring the second voice data based on the adjusted collection state are executed at a local side of a voice device; and

an operation of performing far-field voice recognition on the second voice data using the preset far-field voice recognition model is executed by the voice device at the local side in response to reception of feedback information obtained by an operation of cloud.

12. The apparatus of claim 9 , wherein the processor is configured to:

acquire the first voice data collected by a voice collection device; and

acquire the second voice data collected by a voice collection device after the collection state is adjusted;

wherein the voice collection device comprises: a microphone array provided with more than one microphone for collecting voice data in more than one direction.

13. The apparatus of claim 9 , wherein the processor is configured to perform at least one of the following:

in a case that the voice collection device comprises a microphone array, turning on a microphone, on the location information, in the microphone array; and

in a case that the voice collection device comprises the microphone array, increasing the number of microphones turned on, on the location information, in the microphone array.

14. The apparatus of claim 9 , wherein the processor is configured to perform at least one of the following:

turning off microphones at locations, except the location information, in the microphone array; and

reducing the number of the microphones turned on, at locations, except the location information, in the microphone array.

15. The apparatus of claim 9 , wherein the processor is further configured to:

collect voice data and sound source data of the voice data; and

perform training, after the voice data and the sound source data is preprocessed, with an LSTM model to obtain a far-field voice recognition model based on LSTM.

16. An air conditioner, comprising the voice recognition apparatus of claim 9 .