Voiceprint retrieval method based on deep Hash

Abstract

The invention discloses a voiceprint retrieval method based on deep Hash by which the effects of low storage space and efficient retrieval in a voiceprint retrieval task are achieved. The method comprises a step of training a deep voiceprint hash model, a step of constructing a hash coding database and a step of retrieving the query voice in the database, and is characterized by firstly constructing an end-to-end deep neural network structure, and training the deep neural network model by utilizing the voice data marked with a speaker identity to obtain a deep voiceprint hash function, and then calculating the Hash codes corresponding to the training set through the deep voiceprint Hash function, and constructing a database; for the newly inputted voice data, using the deep voiceprint hashfunction to calculate a corresponding hash code, and adding the hash code to a database in real time. During the retrieval process, for the given voice, the deep voiceprint hash function is used forcalculating the corresponding hash code, and finally a retrieval result is obtained in the database based on index or Hamming distance sorting.

Description

technical field [0001] The invention relates to a voiceprint retrieval method based on deep hashing, which is used to realize fast voiceprint retrieval of a large-scale voice database under low storage overhead. Background technique [0002] Voiceprint retrieval uses a given voice to retrieve one or more voices from the same speaker as this voice in the database. Due to the popularity of microphone recording devices such as mobile phones, personal computers, etc. in recent years, and the rapid development of network media, a large number of voices and videos have sprung up, and hundreds of hours of videos are uploaded to the cloud every minute. The use of voice retrieval is becoming more and more extensive, such as recommending similar voices through voice retrieval; detecting infringements through voice retrieval; in large-scale voiceprint authentication, too many speakers will lead to slow authentication speed, you can also use Retrieval technology speeds up the authentic...

AI Technical Summary

Problems solved by technology

[0005] Purpose of the invention: The current voiceprint retrieval methods are mainly based on real-valued vectors and based on hash coding: the voiceprint retrieval based on real-valued vectors has the problem of low retrieval efficiency in the face of large-scale data; Since the voiceprint retrieval of the Greek code adopts a two-stage training process, the i-vector is extracted first, and then the hash function is used to solve the hash code for the i-vector. The performance of the hash code is constrained by the i-vector. stage training process can not learn more discriminative hash code

[0006] Technical solution: A voiceprint retrieval method based on deep hashing, which introduces an end-to-end deep hashing method, aiming to solve the problem of low retrieval efficiency of the traditional real-valued retrieval system and low accuracy of the two-stage training process of the traditional voiceprint hashing method question

Images