Abstract: The present application discloses a speech recognition method, apparatus, device and readable storage medium, and relates to the technical field of artificial intelligence. A specific implementation includes: an electronic device recognizes a speech signal to obtain a first text; if a first pinyin sequence corresponding to the first text exists in the database, the electronic device uses a correct text corresponding to the first pinyin sequence as the speech recognition result; otherwise, the electronic device performs a fuzzy matching on the first pinyin sequence to obtain multiple second pinyin sequences and second texts corresponding to the second pinyin sequences, and selects the speech recognition result from the multiple second texts.

Abstract: The present disclosure discloses a noise reduction method and apparatus based on in-vehicle sound zones and a medium. The method includes: determining, based on voice signals obtained by sound zones, a target sound zone to which a voice signal waking up a vehicle-mounted terminal belongs, in which each sound zone has a corresponding noise reduction module; and sending a noise reduction termination instruction to noise reduction modules of sound zones other than the target sound zone, in which the noise reduction termination instruction is configured to instruct the noise reduction modules of the sound zones other than the target sound zone to terminate noise reduction algorithms in a process of performing voice interaction for the target sound zone.

Abstract: The present application discloses a voice data processing method, an apparatus, a device and a storage media. The specific implementation solution is: starting a microphone management thread to collect audio data acquired by a microphone in a process of a voice interactive application, and when the voice interactive application is in a non-wake-up state, starting a wake-up thread to perform wake-up processing on the voice interactive application according to the audio data, there is no need to request the microphone separately by a recognition engine which performs voice recognition on the audio data collected by the microphone management thread, thereby realizing the wake-up engine and the recognition engine in the same process. The wake-up engine and the recognition engine avoid a problem of losing part of data due to waiting for preparation of the microphone, and improve efficiency and accuracy of the voice interactive application.

Abstract: The present disclosure discloses a noise reduction method and apparatus based on in-vehicle sound zones and a medium. The method includes: determining, based on voice signals obtained by sound zones, a target sound zone to which a voice signal waking up a vehicle-mounted terminal belongs, in which each sound zone has a corresponding noise reduction module; and sending a noise reduction termination instruction to noise reduction modules of sound zones other than the target sound zone, in which the noise reduction termination instruction is configured to instruct the noise reduction modules of the sound zones other than the target sound zone to terminate noise reduction algorithms in a process of performing voice interaction for the target sound zone.

Abstract: The present application discloses a speech recognition method, apparatus, device and readable storage medium, and relates to the technical field of artificial intelligence. A specific implementation includes: an electronic device recognizes a speech signal to obtain a first text; if a first pinyin sequence corresponding to the first text exists in the database, the electronic device uses a correct text corresponding to the first pinyin sequence as the speech recognition result; otherwise, the electronic device performs a fuzzy matching on the first pinyin sequence to obtain multiple second pinyin sequences and second texts corresponding to the second pinyin sequences, and selects the speech recognition result from the multiple second texts.

Abstract: A computer-implemented method, an apparatus, and an article of manufacture for making a business process management (BPM) application available to multiple tenants in a multi-tenancy software model. The method includes: obtaining a service level agreement (SLA) as required by a tenant in a multi-tenancy software model, in response to a registration request for the BPM software application from the tenant; selecting a multi-tenancy model based on the obtained SLA; calling a predetermined conversion process corresponding to the selected multi-tenancy software model and disposing BPM software components in the BPM software application, such that the BPM application is made available to the tenant as a multi-tenancy BPM software application; and saving in metadata of the tenant, a result of disposing the BPM software components by the conversion process.