Unlock instant, AI-driven research and patent intelligence for your innovation.

VoIP voice recognition method, device, computer equipment and storage medium

A technology of Internet telephony and speech recognition, which is applied in speech recognition, speech analysis, instruments, etc. It can solve problems such as no pauses, high requirements for speakers, invalid speech, etc., and achieve the effect of accurate speech sentence punctuation

Active Publication Date: 2020-11-10
BAIDU ONLINE NETWORK TECH (BEIJIBG) CO LTD
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, in the above-mentioned existing technical solutions, the disadvantages of energy-based speech sentence segmentation include: it is impossible to filter noise and invalid speech, and the requirements for speakers are relatively high, and there must be no pause in the middle
But usually the voice quality during IP telephony is up and down, resulting in intermittent voice

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
  • VoIP voice recognition method, device, computer equipment and storage medium
  • VoIP voice recognition method, device, computer equipment and storage medium
  • VoIP voice recognition method, device, computer equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0030] figure 2 The flow chart of the voice recognition method for Internet phone provided by Embodiment 1 of the present invention, this embodiment is applicable to the situation of sentence segmentation in the voice of Internet phone, the method can be executed by the voice recognition device for Internet phone, and the device can use software and / or hardware implementation. like figure 2 As shown, the VoIP voice recognition method includes:

[0031] Step 110: Determine the energy sentence segmentation probability of the Internet phone voice, and determine candidate sentence break points in the Internet phone voice based on the energy sentence segmentation probability.

[0032] Specifically, after the IP telephone voice is obtained, the energy sentence segmentation probabilities corresponding to each position of the IP telephone voice may be determined, and the candidate sentence break points contained in the IP telephone voice may be obtained according to the energy se...

Embodiment 2

[0052] image 3 It is a flow chart of the voice recognition method for the Internet phone provided by Embodiment 2 of the present invention. On the basis of the first embodiment above, the embodiment of the present invention performs sentence segmentation processing on the voice of the Internet phone according to the screening results to obtain the voice of the Internet phone Steps are added after the voice clauses included: screen out single-person long clauses from the voice clauses according to the preset voice single-sentence length threshold; Sentence correction processing for single long clauses. like image 3 As shown, the VoIP voice recognition method includes:

[0053] Step 210: Determine the energy sentence segmentation probability of the Internet phone voice, and determine candidate sentence break points in the Internet phone voice based on the energy sentence segmentation probability.

[0054] Step 220 , determining the probability that the Internet phone voices...

Embodiment 3

[0078] Figure 4 It is a schematic structural diagram of an Internet phone voice recognition device provided in Embodiment 3 of the present invention. The device executes the Internet phone voice recognition method provided in any one of the above embodiments, and the device can be realized by software and / or hardware. like Figure 4 As shown, the VoIP speech recognition device includes:

[0079] The candidate sentence-breaking point acquisition module 310 is configured to determine the energy sentence-breaking probability of the Internet telephone voice, and determine the candidate sentence-breaking point in the Internet phone voice based on the energy sentence-breaking probability.

[0080] The voice attribution detection module 320 is used to determine the probability that the voice of the Internet phone at the time before and after the candidate break point belongs to different speakers.

[0081] The voice sentence segmentation probability determination module 330 is con...

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

No PUM Login to View More

Abstract

The embodiment of the invention discloses a network telephone voice recognition method, a network telephone voice recognition device, a computer device and a storage medium. The method comprises the following steps: determining the energy sentence-segmenting probability of network telephone voice, and determining candidate sentence-segmenting points in the network telephone voice based on the energy sentence-segmenting probability; determining the probability that the network telephone voice at the moments before and after the candidate sentence-segmenting points belongs to different speakers;according to the energy sentence-segmenting probability of the candidate sentence-segmenting points and the probability of the different speakers, determining the voice sentence-segmenting probability of the candidate sentence-segmenting points; and screening the candidate sentence-segmenting points based on the voice sentence-segmenting probability of the candidate sentence-segmenting points, and carrying out sentence-segmenting treatment on the network telephone voice according to the screening result, thus obtaining voice sub sentences contained in the network telephone voice. With the technical scheme provided by the invention, the problem that for the traditional energy sentence-segmenting method, the accuracy rate in voice sentence segmenting is low, consequently, the voice recognition accuracy rate is not high, is solved, and the effect of accurate voice sentence segmenting of the network telephone voice is realized.

Description

technical field [0001] Embodiments of the present invention relate to voice recognition and voice processing technologies, and in particular to a voice recognition method, device, computer equipment and storage medium for Internet telephony. Background technique [0002] With the rapid development of the communication industry, IP telephony (Voice Over Internet Protocol, Internet telephony) has become a communication method commonly used by the public, and the speech recognition technology in the IP telephony process has also become very important, especially the speech sentence recognition technology. [0003] The current speech recognition process is: speech signal preprocessing → speech segmentation → speech recognition. Speech preprocessing includes speech decoding and denoising, etc. Speech segmentation splits continuous speech into sentence fragments. Speech recognition uses feature extraction, acoustic models, Language models and decoders, etc. Among them, the 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
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G10L15/04G10L17/02G10L17/22
Inventor 岑敏强
Owner BAIDU ONLINE NETWORK TECH (BEIJIBG) CO LTD