Unlock instant, AI-driven research and patent intelligence for your innovation.
Voice separation method and system, mobile terminal and storage medium
What is Al technical title?
Al technical title is built by PatSnap Al team. It summarizes the technical point description of the patent document.
A speech separation and audio technology, applied in speech analysis, neural learning methods, instruments, etc., can solve the problems of poor speech separation effect, achieve the effect of improving accuracy, simplifying steps, and improving user experience
Active Publication Date: 2020-07-03
XIAMEN KUAISHANGTONG TECH CORP LTD
View PDF8 Cites 11 Cited by
Summary
Abstract
Description
Claims
Application Information
AI Technical Summary
This helps you quickly interpret patents by identifying the three key elements:
Problems solved by technology
Method used
Benefits of technology
Problems solved by technology
[0005] The purpose of the embodiment of the present invention is to provide a voice separation method, system, mobile terminal and storage medium, aiming at solving the voice separation effect in the existing voice separation process bad question
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more
Image
Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
Click on the blue label to locate the original text in one second.
Reading with bidirectional positioning of images and text.
Smart Image
Examples
Experimental program
Comparison scheme
Effect test
Embodiment 1
[0044] see figure 1 , is a flowchart of the speech separation method provided by the first embodiment of the present invention, including steps:
[0045] Step S10, obtaining the sample audio of the left channel and the sample audio of the right channel, and combining the sample audio of the left channel and the sample audio of the right channel to obtain the combined sample audio;
[0046] Among them, the left channel sample audio and the right channel sample audio are obtained by collecting the phone audio in the real scene. In this step, the phone call audio is saved through the phone recording function, but the saved call audio is required to be double-sound channel, that is, the left channel is one person's call, and the right channel is another person's call, and 10,000 call audios are recorded, and each call audio is about 2 minutes;
[0047] Specifically, in this step, all the two-channel call audios are merged into single-channel audio, that is, the respective calls o...
Embodiment 2
[0066] see figure 2 , is a flow chart of the speech separation method provided by the second embodiment of the present invention, including steps:
[0067] Step S11, obtaining the sample audio of the left channel and the sample audio of the right channel, and combining the sample audio of the left channel and the sample audio of the right channel to obtain the combined sample audio;
[0068] Step S21, constructing a prenet network, and constructing a CBHG network after the prenet network;
[0069] Wherein, the prenet network includes three fully connected layers, and the CBHG network includes a first convolutional layer, a pooling layer, a second convolutional layer, and a third convolutional layer in sequence, and the second convolutional layer and The third convolutional layer is a one-dimensional convolutional layer, and the filter size of the second convolutional layer and the third convolutional layer is 3, the stride is 1, and the activation function used by the second...
Embodiment 3
[0093] see image 3 , is a schematic structural diagram of the speech separation system 100 provided by the third embodiment of the present invention, including: a sample audio acquisition module 10, a feature dimensionality reduction module 11, a feature decoding module 12, an iterative training module 13 and a speech separation module 14, wherein:
[0094] The sample audio acquisition module 10 is configured to acquire the left channel sample audio and the right channel sample audio, and combine the left channel sample audio and the right channel sample audio to obtain combined sample audio.
[0095] The feature dimensionality reduction module 11 is configured to construct an encoding network, and input the spectral features of the combined sample audio into the decoding network for dimensionality reduction encoding to obtain dimensionality reduction features.
[0096] Wherein, the feature dimensionality reduction module 11 is also used to: construct a prenet network, and co...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More
PUM
Login to View More
Abstract
The invention provides a voice separation method and system, a mobile terminal and a storage medium, wherein the method comprises the steps: obtaining a left channel sample audio and a right channel sample audio, combining the left channel sample audio and the right channel sample audio, and obtaining a combined sample audio; constructing a coding network, and inputting the frequency spectrum features of the combined frequency spectrum into a decoding network for dimension reduction coding to obtain dimension reduction features; performing attention calculation on the dimension reduction features by adopting an attention mechanism to obtain an attention probability value, and inputting the attention probability value into the decoding network for decoding to obtain a frequency spectrum decoding result; calculating a loss value between the frequency spectrum decoding result and the frequency spectrum characteristics of the sample audio, and performing model iterative training on the coding network and the decoding network according to the loss value to obtain a voice separation model; and inputting the voice to be recognized into the voice separation model for voice separation to obtain a left channel audio file and a right channel audio file. According to the invention, the voice separation effect in the voice data is realized by adopting an end-to-end model, and the voice separation accuracy is improved.
Description
technical field [0001] The invention belongs to the technical field of voice separation, and in particular relates to a voice separation method, system, mobile terminal and storage medium. Background technique [0002] Now more and more people communicate and communicate by phone, but the voices of the two parties on the phone are usually combined in the same audio channel, so it is necessary to extract the audio of the two people from the single channel separately for convenience. Subsequent corresponding voice recognition and voiceprint recognition. [0003] The existing voice separation method is to divide the current entire audio segment into multiple independent audio segments by segmenting the silent segment in the voice, and then cluster all the audio segments. The number of clustering categories is two, and the clustering After the class is completed, the corresponding audio clips of the two types of audio are spliced into a complete audio, so as to perform speech...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More
Application Information
Patent Timeline
Application Date:The date an application was filed.
Publication Date:The date a patent or application was officially published.
First Publication Date:The earliest publication date of a patent with the same application number.
Issue Date:Publication date of the patent grant document.
PCT Entry Date:The Entry date of PCT National Phase.
Estimated Expiry Date:The statutory expiry date of a patent right according to the Patent Law, and it is the longest term of protection that the patent right can achieve without the termination of the patent right due to other reasons(Term extension factor has been taken into account ).
Invalid Date:Actual expiry date is based on effective date or publication date of legal transaction data of invalid patent.
Login to View More 
Login to View More