WO2018129869A1 - 声纹验证方法和装置 - Google Patents

声纹验证方法和装置 Download PDF

Info

Publication number
WO2018129869A1
WO2018129869A1 PCT/CN2017/090171 CN2017090171W WO2018129869A1 WO 2018129869 A1 WO2018129869 A1 WO 2018129869A1 CN 2017090171 W CN2017090171 W CN 2017090171W WO 2018129869 A1 WO2018129869 A1 WO 2018129869A1
Authority
WO
WIPO (PCT)
Prior art keywords
voiceprint
feature
waveform
matching
sound waveform
Prior art date
Application number
PCT/CN2017/090171
Other languages
English (en)
French (fr)
Inventor
柳岸
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2018129869A1 publication Critical patent/WO2018129869A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/30Authentication, i.e. establishing the identity or authorisation of security principals
    • G06F21/31User authentication
    • G06F21/32User authentication using biometric data, e.g. fingerprints, iris scans or voiceprints
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/06Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/08Speech classification or search
    • G10L15/10Speech classification or search using distance or distortion measures between unknown speech and reference templates
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L17/00Speaker identification or verification techniques
    • G10L17/04Training, enrolment or model building

Definitions

  • the present disclosure relates to the field of identity verification technologies, for example, to a voiceprint verification method and apparatus.
  • voiceprint recognition With the gradual maturity of speech technology, speech recognition, text to speech (TTS) synthesis, language recognition, voiceprint recognition and other technologies have been applied to the field of speech. Since the voiceprint is unique, after the voiceprint model is established, the user input voiceprint is matched with the voiceprint model, and the user identity can be verified in this way.
  • voiceprint Although the voiceprint is unique, it cannot avoid malicious simulation. For example, by simulating the user's voiceprint through recording and other means, it can successfully pass the authentication. Therefore, using voiceprint for identity verification has security risks.
  • a voiceprint verification method and device can solve the problem of maliciously simulating a user's voiceprint through identity verification.
  • a voiceprint verification method comprising:
  • the method before the collecting the voice information to be verified, the method further includes:
  • the sound waveform of the speech segment is stored as the pre-stored standard sound waveform.
  • the method before the collecting the voice information to be verified, the method further includes:
  • the collecting the voice information to be verified includes:
  • the random password input by the user by voice is collected as voice information to be verified.
  • the sound waveform is matched with the pre-stored standard sound waveform, and the voiceprint feature and the pre-stored standard voiceprint feature are feature-matched, including:
  • the voiceprint verification fails; if the waveform matching is successful, the voiceprint feature and the pre-stored standard voiceprint are Feature matching is performed, if the feature matching is successful, the voiceprint verification is passed, and if the feature matching is unsuccessful, the voiceprint verification fails;
  • the voiceprint verification fails; if the feature matching is successful, the sound waveform and the pre-stored standard sound are The waveform performs waveform matching. If the waveform matching is successful, the voiceprint verification passes, and if the waveform matching is unsuccessful, the voiceprint verification fails.
  • extracting voiceprint features including:
  • the sound waveform is converted to an acoustic spectrum in which the voiceprint features are extracted.
  • a voiceprint verification device comprising:
  • the acquisition module is configured to collect voice information to be verified
  • Extracting a module configured to extract a voiceprint feature in a sound waveform corresponding to the voice information
  • a verification module configured to perform waveform matching on the sound waveform and the pre-stored standard sound waveform, to perform feature matching on the voiceprint feature and the pre-stored standard voiceprint feature; and if the waveform match and the feature match match If successful, the voiceprint verification will pass.
  • the collecting module is further configured to:
  • the sound waveform of the speech segment is stored as the pre-stored standard sound waveform.
  • the collecting module is further configured to: before the collecting the voice information to be verified, generate a random password according to the intercepted voice segment and store the random password.
  • the collection module is configured to:
  • the random password input by the user by voice is collected as voice information to be verified.
  • the verification module is set to:
  • the voiceprint verification fails; if the waveform matching is successful, the voiceprint feature and the pre-stored standard voiceprint are Feature matching is performed, if the feature matching is successful, the voiceprint verification is passed, and if the feature matching is unsuccessful, the voiceprint verification fails;
  • the voiceprint verification fails; if the feature matching is successful, the sound waveform and the pre-stored standard sound are The waveform performs waveform matching. If the waveform matching is successful, the voiceprint verification passes, and if the waveform matching is unsuccessful, the voiceprint verification fails.
  • extracting voiceprint features including:
  • the sound waveform is converted to an acoustic spectrum in which the voiceprint features are extracted.
  • a computer readable storage medium storing computer executable instructions arranged to perform the above method.
  • a terminal device comprising:
  • At least one processor At least one processor
  • the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the method described above.
  • the above technical solution not only performs matching verification on the voiceprint feature, but also performs matching verification on the sound waveform, and both matching verifications are passed, and the voiceprint verification is confirmed. In this way, even if the voiceprint feature of the user is maliciously simulated, the voiceprint feature and the sound waveform are not simulated at the same time, and the above technical solution avoids the problem that the user's voiceprint feature is maliciously simulated and can pass the identity verification.
  • 1 is a flow chart of a voiceprint verification method of the first embodiment
  • Figure 2 is a flow chart of the voiceprint verification method of the second embodiment
  • Figure 3 is a structural view of a voiceprint verification device of a third embodiment
  • FIG. 4 is a schematic diagram showing the hardware structure of a terminal device according to an embodiment.
  • This embodiment provides a voiceprint verification method.
  • 1 is a flow chart of a voiceprint verification method of the first embodiment.
  • the execution body of this embodiment may be a terminal device.
  • step 110 voice information to be verified is collected.
  • the terminal device can collect voice information input by the user, and the voice information is voice information to be verified.
  • the voice information may be a voice password input by the user through voice.
  • the voiceprint can be a voice feature contained in the voice that can characterize and identify the speaker.
  • the voiceprint feature may be a parameter extracted from the speaker's voice that characterizes the personality of the speaker's voice. Collecting voice information can include:
  • the voiceprint verification function is activated to prompt the user voice to input the voice password by voice, wherein the voice password can be a piece of text or a number, and the user can read the piece of text or numbers;
  • the voice password input by the user is collected through the microphone of the terminal device.
  • the voice signal is the carrier of the voice information
  • the voice signal is the sound with the waveform
  • the voice password read by the user is carried in In the sound waveform.
  • step 120 the voiceprint feature is extracted in the sound waveform corresponding to the voice information.
  • the sound waveform may be a serial bit stream in binary representation in the terminal device, and the sound waveform carries the waveform of the voice information input by the user.
  • the same voice password is input, since the voices of multiple users may be different, the speaking modes may be different, resulting in different voice waveforms of multiple users.
  • the sound waveform can be converted into a sound wave spectrum by Fourier transform, and the voiceprint feature can be extracted in the sound wave spectrum.
  • the voiceprint feature can comprehensively characterize the wavelength, frequency, intensity, and rhythm of the sound.
  • Each user's voiceprint features are unique.
  • the voiceprint feature may be a Linear Prediction Coefficient (LPC), a Perceptual Linear Predictive (PLP) coefficient, or a Mel-frequency Cepstrum Coefficient (MFCC).
  • LPC Linear Prediction Coefficient
  • PDP Perceptual Linear Predictive
  • MFCC Mel-frequency Cepstrum Coefficient
  • step 130 the sound waveform is matched with the pre-stored standard sound waveform, and the voiceprint feature and the pre-stored standard voiceprint feature are feature-matched.
  • the standard sound waveform may be a sound waveform of voice information previously input by a legitimate user of the terminal device.
  • the voice segment input by the user may be intercepted; the sound waveform of the intercepted voice segment is stored as a standard sound waveform; and a random password is generated according to the intercepted voice segment and the random password (voice password) is stored.
  • the speech segment may be part of the speech information intercepted in the speech information input by the user. For example, the user inputs the voice information “Good weather today”, and part of the voice information intercepted in the voice information may be “good weather”, and the “good weather” is a voice segment.
  • the random password may be text information formed by speech recognition of the speech segment.
  • the pre-stored random password may be obtained; the user is prompted to input the obtained random password by using a voice mode; and the random password input by the user by using the voice mode may be collected, and the collected random number may be collected.
  • the password is used as the voice information to be verified.
  • the standard voiceprint feature legalizes the voiceprint characteristics of the user.
  • the voice information input by the legal user may be collected in advance, and the voiceprint feature of the legitimate user is extracted according to the voice information and stored as a standard voiceprint feature.
  • Waveform matching and feature matching can be performed simultaneously or sequentially. When matching is performed in order, waveform matching can be performed first, and then feature matching can be performed; feature matching can also be performed first, and then waveform matching can be performed.
  • the waveform matching may be to calculate the similarity between the sound waveform of the voice signal input by the user and the standard sound waveform. If the similarity of the waveform is greater than the preset waveform similarity threshold, the waveform is determined to be matched if the similarity of the waveform is less than or equal to the preset.
  • the waveform similarity threshold determines that the waveforms do not match.
  • the waveform similarity threshold is an empirical value or an experimentally obtained value, for example, 98%.
  • the feature matching may be to calculate the similarity between the voiceprint feature of the voice signal input by the user and the standard voiceprint feature. If the similarity of the feature is greater than the preset feature similarity threshold, the feature matching is determined if the feature similarity is less than or equal to The preset feature similarity threshold determines that the features do not match.
  • the feature similarity threshold is an empirical value or an experimentally obtained value, for example, 98%.
  • step 140 if both the waveform matching and the feature matching match successfully, the voiceprint verification passes.
  • the voiceprint verification is passed, indicating that the voice information to be verified is legal, and the user who inputs the voice information to be verified is a legitimate user.
  • the voiceprint verification does not pass.
  • the voiceprint verification fails, indicating that the user who inputs the voice information to be verified is an illegal user.
  • the voiceprint feature is matched and verified, but also the sound waveform is matched and verified, and both matching verifications are passed, and the voiceprint verification is confirmed. Even if the voiceprint feature of the user is maliciously simulated, the voiceprint feature and the sound waveform are not simulated at the same time, and the voiceprint feature of the user is prevented from being maliciously simulated, and the security of the identity verification is improved through the phenomenon of identity verification.
  • the sound waveform may be first matched with the pre-stored standard sound waveform. If the waveform matching is unsuccessful, the voiceprint verification does not pass; if the waveform matching is successful, the voiceprint feature and the pre-stored The standard voiceprint feature performs feature matching; if the feature matching is successful, the voiceprint verification passes, and if the feature matching is unsuccessful, the voiceprint verification fails. It is also possible to first match the voiceprint feature with the pre-stored standard voiceprint feature. If the feature matching is unsuccessful, the voiceprint verification does not pass; if the feature matching is successful, the sound waveform and the pre-stored standard sound waveform are waveformd. Matching, if the waveform matching is successful, the voiceprint verification passes, and if the waveform matching is unsuccessful, the voiceprint verification does not pass.
  • Fig. 2 is a flow chart of the voiceprint verification method of the second embodiment.
  • step 210 the user's standard voiceprint features are extracted.
  • the user is prompted to input voice information, and the voice information input by the user is recorded.
  • voice information the voiceprint feature of the user is extracted, and the voiceprint feature of the user is stored in the voiceprint model library.
  • This step 210 can be performed when the terminal device is initialized.
  • step 220 the voice segment input by the user is intercepted.
  • the voice segment input by the user may be intercepted, and the standard sound waveform corresponding to the voice segment and the random password generated according to the voice segment are used in the next voiceprint verification.
  • the user can enter a newly generated random password and use the newly stored standard sound waveform.
  • the voice segment may be intercepted in the voice information used to extract the standard voiceprint feature, a random password is generated according to the voice segment, and the sound waveform of the voice segment is used as a standard sound waveform.
  • step 230 a random password is generated based on the voice segment and a random password is stored, and the sound waveform of the voice segment is stored as a standard sound waveform.
  • the voice information input by the user is recorded; in the recorded voice information, multiple voice segments are intercepted; and the sound waveforms of the plurality of voice segments are used as standard sound waveforms.
  • a random password may be generated according to each of the voice segments; and a plurality of random passwords respectively corresponding to the voice segments are stored.
  • the content of the call is recorded, the voice segment of the user is intercepted, a random password is generated according to the voice segment, and the sound waveform of the voice segment is used as a standard sound waveform.
  • step 240 when performing voiceprint verification, the user is prompted to input a random password corresponding to the voice segment.
  • the user can activate the voiceprint verification function of the terminal device to perform this voiceprint verification.
  • the random password is displayed on the screen, and the user is prompted to input the random password by voice.
  • the voice clip is "Good weather” and you can prompt the user to enter "Good weather”.
  • step 250 the random password input by the user according to the prompt voice is collected to form voice information to be verified.
  • step 260 the sound waveform of the voice information and the standard sound waveform are waveform matched. If the waveform matching is successful, step 270 is performed; if the waveform matching fails, step 290 is performed.
  • step 270 the voiceprint features of the voice information are matched to the standard voiceprint features. If the feature matching is successful, step 280 is performed; if the feature matching fails, step 290 is performed.
  • step 280 the voiceprint verification is passed.
  • step 290 the voiceprint verification does not pass.
  • the voice segment currently required by the user in this embodiment is different from the voice segment that needs to be input by the user.
  • the standard sound waveform currently used is different from the standard sound waveform used in the previous time, and the voice of the user may be determined before the feature matching is performed. Whether the waveform matches the sound waveform of the stored speech segment, and the feature matching is performed on the premise that the waveform matching is successful, thereby improving the security of the user identity verification.
  • FIG. 3 is a structural diagram of a voiceprint verification device of a third embodiment.
  • the device of this embodiment may be disposed in a terminal device.
  • the device includes an acquisition module 310, an extraction module 320, and a verification module 33.
  • the acquisition module 310 is configured to collect voice information to be verified.
  • the extraction module 320 is configured to extract the voiceprint feature in the sound waveform corresponding to the voice information.
  • the verification module 330 is configured to perform waveform matching on the sound waveform and the pre-stored standard sound waveform, to perform feature matching on the voiceprint feature and the pre-stored standard voiceprint feature; and if the waveform matching and the feature matching match If successful, the voiceprint verification will pass.
  • the collecting module 310 may be further configured to: intercept the voice segment input by the user before acquiring the voice information to be verified; and store the sound waveform of the voice segment as the pre-stored standard sound waveform.
  • the collecting module 310 may be further configured to generate a random password according to the intercepted voice segment and store the random password before the collecting the voice information to be verified.
  • the collecting module 310 may be configured to acquire a pre-stored random password; prompt the user to input the obtained random password by using a voice mode; and collect the random password input by the user by using a voice manner as the to-be-verified voice message.
  • the verification module 330 can be configured to: the sound waveform and the pre-stored standard The sound waveform is waveform matched, if the waveform matching is unsuccessful, the voiceprint verification fails; if the waveform matching is successful, the voiceprint feature and the pre-stored standard voiceprint feature are feature-matched if the feature If the matching is successful, the voiceprint verification is passed. If the feature matching is unsuccessful, the voiceprint verification fails; or the verification module 330 may be configured to: perform matching on the voiceprint feature and the pre-stored standard voiceprint feature. If the feature matching is unsuccessful, the voiceprint verification fails; if the feature matching is successful, the sound waveform is matched with the pre-stored standard sound waveform, and if the waveform matching is successful, the voiceprint verification is passed. If the waveform matching is unsuccessful, the voiceprint verification does not pass.
  • the present embodiment provides a computer readable storage medium storing computer executable instructions arranged to perform the method of any of the above embodiments.
  • the terminal device includes:
  • At least one processor 40 one processor 40 is exemplified in FIG. 4; a memory 41, a voice input device (such as a microphone) 42 and a display 43; and a communication interface 44 and a bus 45.
  • the processor 40, the memory 41, the voice input device (such as a microphone) 42, the display 43 and the communication interface 44 can complete communication with each other through the bus 45.
  • a voice input device (such as a microphone) 42 can collect voice information.
  • Display 43 can display a random password for the user to read.
  • the communication interface 44 can receive signals as well as voice data, image data, or video data.
  • Processor 40 may invoke logic instructions in memory 41 to perform the methods of the above-described embodiments.
  • logic instructions in the memory 41 described above may be implemented in the form of a software functional unit and sold or used as a stand-alone product, and may be stored in a computer readable storage medium.
  • the memory 41 is a computer readable storage medium and can be used to store a software program, a computer executable program, such as a program instruction or a module corresponding to the method in the above embodiment.
  • the processor 40 executes the functional application and data processing by executing software programs, instructions or modules stored in the memory 41, i.e., implements the methods in the above embodiments.
  • the memory 41 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application required for at least one function; and the storage data area may be stored according to the use of the terminal device The data created, etc. Further, the memory 41 may include a high speed random access memory, and may also include a nonvolatile memory.
  • the above technical solution may be embodied in the form of a software product stored in a storage medium, including one or more instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to execute All or part of the steps of the method described in the above embodiments.
  • the foregoing storage medium may be a non-transitory storage medium, including: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
  • a voiceprint verification method and device prevent the user's voiceprint feature from being maliciously simulated, and the phenomenon of identity verification improves the security of voiceprint verification.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Computer Security & Cryptography (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • Telephone Function (AREA)
  • Telephonic Communication Services (AREA)

Abstract

一种声纹验证方法和装置,其中,该方法包括:采集待验证的语音信息;在语音信息对应的声音波形中,提取声纹特征;将声音波形和预存的标准声音波形进行波形匹配,将声纹特征和预存的标准声纹特征进行特征匹配;以及如果波形匹配和特征匹配都匹配成功,则声纹验证通过。

Description

声纹验证方法和装置 技术领域
本公开涉及身份验证技术领域,例如涉及一种声纹验证方法和装置。
背景技术
随着语音技术的逐渐成熟,语音识别、从文本到语言(Text To Speech,TTS)合成、语种识别、声纹识别等多项技术被应用到语音领域。由于声纹具有唯一性,在建立声纹模型后,将用户输入的声纹与该声纹模型进行匹配,可以通过这种方式验证用户身份。
声纹虽然具有唯一性,但无法避免恶意模拟,比如:通过录音等手段模拟用户的声纹,可以成功通过身份验证,因此采用声纹进行身份验证存在安全隐患。
发明内容
一种声纹验证方法和装置,可以解决恶意模拟用户的声纹,通过身份验证的问题。
一种声纹验证方法,包括:
采集待验证的语音信息;
在所述语音信息对应的声音波形中,提取声纹特征;
将所述声音波形和预存的标准声音波形进行波形匹配,将所述声纹特征和预存的标准声纹特征进行特征匹配;以及
如果所述波形匹配和所述特征匹配都匹配成功,则声纹验证通过。
可选的,在所述采集待验证的语音信息之前,所述方法还包括:
截取用户输入的语音片段;以及
将所述语音片段的声音波形存储为所述预存的标准声音波形。
可选的,在所述采集待验证的语音信息之前,所述方法还包括:
根据截取的所述语音片段生成随机密码并存储所述随机密码。
可选的,所述采集待验证的语音信息,包括:
获取预先存储的随机密码;
提示用户通过语音方式输入获取的所述随机密码;以及
采集所述用户通过语音方式输入的所述随机密码,作为待验证的语音信息。
可选的,将所述声音波形和预存的标准声音波形进行波形匹配,将所述声纹特征和预存的标准声纹特征进行特征匹配,包括:
将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配不成功,则声纹验证不通过;如果所述波形匹配成功,则将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配成功,则声纹验证通过,如果所述特征匹配不成功,则声纹验证不通过;
或者,
将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配不成功,则声纹验证不通过;如果所述特征匹配成功,则将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配成功,则声纹验证通过,如果所述波形匹配不成功,则声纹验证不通过。
可选的,在所述语音信息对应的声音波形中,提取声纹特征,包括:
将所述声音波形转换为声波频谱,在所述声波频谱中提取声纹特征。
一种声纹验证装置,包括:
采集模块,设置为采集待验证的语音信息;
提取模块,设置为在所述语音信息对应的声音波形中,提取声纹特征;以及
验证模块,设置为将所述声音波形和预存的标准声音波形进行波形匹配,将所述声纹特征和预存的标准声纹特征进行特征匹配;以及如果所述波形匹配和所述特征匹配都匹配成功,则声纹验证通过。
可选的,所述采集模块,还设置为:
在采集待验证的语音信息之前,截取用户输入的语音片段;以及
将所述语音片段的声音波形存储为所述预存的标准声音波形。
可选的,所述采集模块,还设置为:在所述采集待验证的语音信息之前,根据截取的所述语音片段生成随机密码并存储所述随机密码。
可选的,所述采集模块设置为:
获取预先存储的随机密码;
提示用户通过语音方式输入获取的所述随机密码;以及
采集所述用户通过语音方式输入的所述随机密码,作为待验证的语音信息。
可选的,所述验证模块设置为:
将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配不成功,则声纹验证不通过;如果所述波形匹配成功,则将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配成功,则声纹验证通过,如果所述特征匹配不成功,则声纹验证不通过;
或者,
将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配不成功,则声纹验证不通过;如果所述特征匹配成功,则将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配成功,则声纹验证通过,如果所述波形匹配不成功,则声纹验证不通过。
可选的,在所述语音信息对应的声音波形中,提取声纹特征,包括:
将所述声音波形转换为声波频谱,在所述声波频谱中提取声纹特征。
一种计算机可读存储介质,存储有计算机可执行指令,所述计算机可执行指令设置为执行上述方法。
一种终端设备,包括:
至少一个处理器;以及
与所述至少一个处理器通信连接的存储器;其中,
所述存储器存储有可被所述至少一个处理器执行的指令,所述指令被所述至少一个处理器执行,以使所述至少一个处理器执行上述的方法。
以上技术方案不但对声纹特征进行匹配验证,还要对声音波形进行匹配验证,两个匹配验证都通过,才认定声纹验证通过。这样即便用户的声纹特征被恶意模拟,也不会发生声纹特征和声音波形同时被模拟的情况,进而通过以上技术方案避免了用户的声纹特征被恶意模拟,可以通过身份验证的问题。
附图说明
图1是第一实施例的声纹验证方法的流程图;
图2是第二实施例的声纹验证方法的流程图;
图3是第三实施例的声纹验证装置的结构图;以及
图4是一实施例的终端设备的硬件结构示意图。
具体实施方式
以下结合附图以及实施例,对以下技术方案进行详细说明。此处所描述的实施例仅仅用以解释技术方案。在不冲突的情况下,以下实施例以及实施例中的技术特征可以相互任意组合。
实施例一
本实施例提供一种声纹验证方法。图1是第一实施例的声纹验证方法的流程图。本实施例的执行主体可以为终端设备。
在步骤110中,采集待验证的语音信息。
在启动终端设备的声纹验证功能后,终端设备可以采集用户输入的语音信息,该语音信息为待验证的语音信息。在本实施例中,该语音信息可以是用户通过语音输入的语音密码。声纹可以是语音中蕴含的、能表征和标识说话人的语音特征。声纹特征可以是从说话人的语音中提取出来的、可以表征该说话人语音的个性特征的参数。采集语音信息可以包括:
启动声纹验证功能,提示用户语音通过语音输入语音密码,其中,语音密码可以是一段文字或数字,用户可以读出该段文字或数字;以及
通过终端设备的麦克风(Microphone)采集用户输入的语音密码。语音信号为语音信息的载体,语音信号为具有波形的声音,用户读出的语音密码承载在 声音波形中。
在步骤120中,在该语音信息对应的声音波形中,提取声纹特征。
声音波形在终端设备中可以是采用二进制表示的串行比特流,声音波形承载用户输入的语音信息的波形。在输入同一语音密码时,由于多个用户的音色可能不同,说话方式也可能不同,导致多个用户的声音波形不同。
可以通过傅里叶变换将声音波形转换为声波频谱,在声波频谱中提取声纹特征。声纹特征可以综合表征声音的波长、频率、强度和节奏。每个用户的声纹特征具有唯一性。所述声纹特征可以是线性预测系数(Linear Prediction Coefficient,LPC)、感知线性预测(Perceptual Linear Predictive,PLP)系数或者梅尔倒谱系数(Mel-frequency Cepstrum Coefficient,MFCC)。
在步骤130中,将该声音波形和预存的标准声音波形进行波形匹配,将该声纹特征和预存的标准声纹特征进行特征匹配。
标准声音波形可以为使用终端设备的合法用户预先输入的语音信息的声音波形。
可以截取用户输入的语音片段;将截取的语音片段的声音波形作为标准声音波形进行存储;以及根据截取的语音片段生成随机密码并存储所述随机密码(语音密码)。语音片段可以是在用户输入的语音信息中截取的部分语音信息。例如:用户输入语音信息“今天天气不错”,在该语音信息中截取的部分语音信息可以是“天气不错”,该“天气不错”即是语音片段。可选地,随机密码可以是对语音片段进行语音识别而形成的文本信息。
在采集待验证的语音信息时,可以获取预先存储的随机密码;提示用户通过语音方式输入获取的所述随机密码;以及采集所述用户通过语音方式输入的所述随机密码,可以将采集的随机密码作为待验证的语音信息。
标准声纹特征合法用户的声纹特征。可以预先采集合法用户输入的语音信息,根据该语音信息提取出合法用户的声纹特征并存储为标准声纹特征。
波形匹配和特征匹配可以同时进行,或者按先后顺序进行。按先后顺序进行匹配时,可以先进行波形匹配,再进行特征匹配;也可以先进行特征匹配,再进行波形匹配。
波形匹配可以是计算用户输入的语音信号的声音波形和标准声音波形的相似度,如果波形的相似度大于预设的波形相似度阈值,则认定波形匹配,如果波形的相似度小于或等于预设的波形相似度阈值,则认定波形不匹配。波形相似度阈值为经验值或实验获得的值,例如为98%。
特征匹配可以是计算用户输入的语音信号的声纹特征和标准声纹特征的相似度,如果特征的相似度大于预设的特征相似度阈值,则认定特征匹配,如果特征的相似度小于或等于预设的特征相似度阈值,则认定特征不匹配。特征相似度阈值为经验值或实验获得的值,例如为98%。
在步骤140中,如果波形匹配和特征匹配都匹配成功,则声纹验证通过。
声纹验证通过,说明待验证的语音信息合法,输入该待验证的语音信息的用户为合法用户。
如果波形匹配和特征匹配中的一个或两个匹配失败,则声纹验证不通过。声纹验证不通过,说明输入待验证的语音信息的用户为非法用户。
本实施例不但对声纹特征进行匹配验证,还要对声音波形进行匹配验证,两个匹配验证都通过,才认定声纹验证通过。即便用户的声纹特征被恶意模拟,也不会发生声纹特征和声音波形同时被模拟的情况,避免了用户的声纹特征被恶意模拟,通过身份验证的现象,提升身份验证的安全性。
实施例二
在本实施例中,可以先将声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配不成功,则声纹验证不通过;如果波形匹配成功,则再将声纹特征和预存的标准声纹特征进行特征匹配;如果特征匹配成功,则声纹验证通过,如果所述特征匹配不成功,则声纹验证不通过。也可以先将声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配不成功,则声纹验证不通过;如果特征匹配成功,则将声音波形和预存的标准声音波形进行波形匹配,如果波形匹配成功,则声纹验证通过,如果所述波形匹配不成功,则声纹验证不通过。
图2是第二实施例的声纹验证方法的流程图。
在步骤210中,提取用户的标准声纹特征。
提示用户输入语音信息,录制用户输入的语音信息,在该语音信息中,提取用户的声纹特征,将用户的声纹特征存储到声纹模型库中。
该步骤210可以在终端设备初始化时进行。
在步骤220中,截取用户输入的语音片段。
为了提升声纹验证的安全性,可以在每次声纹验证通过之后,截取用户输入的语音片段,将该语音片段对应的标准声音波形以及根据语音片段生成的随机密码用在下一次声纹验证中,每次进行声纹验证,用户可以输入新生成的随机密码,使用新存储的标准声音波形。在首次进行声纹验证时,可以在提取标准声纹特征时所使用的语音信息中,截取语音片段,根据该语音片段生成随机密码,并将该语音片段的声音波形作为标准声音波形。
在步骤230中,根据该语音片段生成随机密码并存储随机密码,以及将该语音片段的声音波形作为标准声音波形进行存储。
可选的,在用户使用语音功能的过程中,录制用户输入的语音信息;在录制的所述语音信息中,截取多个语音片段;将多个所述语音片段的声音波形都作为标准声音波形进行存储。可以根据每个所述语音片段生成一个随机密码;并存储多个所述语音片段分别对应的随机密码。
例如:在用户通话过程中,录制通话内容,截取用户的语音片段,根据该语音片段生成随机密码,并将该语音片段的声音波形作为标准声音波形。
在步骤240中,在进行声纹验证时,提示用户语音输入与语音片段对应的随机密码。
用户可以启动终端设备的声纹验证功能进行本次声纹验证。在存储的多个随机密码中,获取其中一个随机密码,在屏幕中显示该随机密码,并提示用户通过语音方式输入该随机密码。例如:语音片段为“天气不错”,可以提示用户语音输入“天气不错”。
在步骤250中,采集该用户根据所述提示语音输入的随机密码,形成待验证的语音信息。
在步骤260中,将该语音信息的声音波形和标准声音波形进行波形匹配。如果波形匹配成功,则执行步骤270;如果波形匹配失败,则执行步骤290。
在步骤270中,将该语音信息的声纹特征和标准声纹特征进行特征匹配。如果特征匹配成功,则执行步骤280;如果特征匹配失败,则执行步骤290。
在步骤280中,声纹验证通过。
在步骤290中,声纹验证不通过。
本实施例的当前需要用户输入的语音片段与前一次需要用户输入的语音片段不同,当前使用的标准声音波形与前一次使用的标准声音波形不同,在进行特征匹配之前,可以先确定用户的声音波形是否和存储的语音片段的声音波形是否匹配,在波形匹配成功的前提下,进行特征匹配,提高了用户身份验证的安全性。
实施例三
本实施例提供一种声纹验证装置。图3是第三实施例的声纹验证装置的结构图。本实施例的所述装置可以设置在终端设备中。
该装置包括:采集模块310、提取模块320以及验证模块33。
采集模块310设置为采集待验证的语音信息。
提取模块320设置为在所述语音信息对应的声音波形中,提取声纹特征。
验证模块330设置为将所述声音波形和预存的标准声音波形进行波形匹配,将所述声纹特征和预存的标准声纹特征进行特征匹配;以及如果所述波形匹配和所述特征匹配都匹配成功,则声纹验证通过。
在一个实施例中,采集模块310,还可以设置为在采集待验证的语音信息之前,截取用户输入的语音片段;以及将所述语音片段的声音波形存储为为所述预存的标准声音波形。
在一实施例中,采集模块310,还可以设置为在所述采集待验证的语音信息之前,根据截取的所述语音片段生成随机密码并存储所述随机密码。
在一实施例中,采集模块310可以设置为获取预先存储的随机密码;提示用户通过语音方式输入获取的所述随机密码;以及采集所述用户通过语音方式输入的所述随机密码作为待验证的语音信息。
在一实施例中,验证模块330可以设置为:将所述声音波形和预存的标准 声音波形进行波形匹配,如果所述波形匹配不成功,则声纹验证不通过;如果所述波形匹配成功,则将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配成功,则声纹验证通过,如果所述特征匹配不成功,则声纹验证不通过;或者,验证模块330可以设置为:将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配不成功,则声纹验证不通过;如果所述特征匹配成功,则将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配成功,则声纹验证通过,如果所述波形匹配不成功,则声纹验证不通过。
本实施例所述的装置的功能在图1~2所示的方法实施例中进行了描述,故本实施例的描述中未详尽之处,可以参见前述实施例中的相关说明。
本实施例提供了一种计算机可读存储介质,存储有计算机可执行指令,所述计算机可执行指令设置为执行上述任一实施例中的方法。
本实施例提供了一种终端设备的硬件结构示意图。参见图4,该终端设备包括:
至少一个处理器(processor)40,图4中以一个处理器40为例;存储器(memory)41、语音输入设备(如麦克风)42以及显示器43;还可以包括通信接口(Communications Interface)44和总线45。其中,处理器40、存储器41、语音输入设备(如麦克风)42、显示器43以及通信接口44可以通过总线45完成相互间的通信。语音输入设备(如麦克风)42可以采集语音信息。显示器43可以显示供用户读取的随机密码。通信接口44可以接收信号,也可以传输语音数据、图像数据或视频数据。处理器40可以调用存储器41中的逻辑指令,以执行上述实施例的方法。
此外,上述的存储器41中的逻辑指令可以通过软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中。
存储器41作为一种计算机可读存储介质,可用于存储软件程序、计算机可执行程序,如上述实施例中的方法对应的程序指令或模块。处理器40通过运行存储在存储器41中的软件程序、指令或模块,从而执行功能应用以及数据处理,即实现上述实施例中的方法。
存储器41可包括存储程序区和存储数据区,其中,存储程序区可存储操作系统、至少一个功能所需的应用程序;存储数据区可存储根据终端设备的使用 所创建的数据等。此外,存储器41可以包括高速随机存取存储器,还可以包括非易失性存储器。
以上技术方案可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括一个或多个指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行上述实施例所述方法的全部或部分步骤。而前述的存储介质可以是非暂态存储介质,包括:U盘、移动硬盘、只读存储器(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等多种可以存储程序代码的介质,也可以是暂态存储介质。
工业实用性
一种声纹验证方法和装置,避免了用户的声纹特征被恶意模拟时,通过身份验证的现象,提高了声纹验证的安全性。

Claims (13)

  1. 一种声纹验证方法,包括:
    采集待验证的语音信息;
    在所述语音信息对应的声音波形中,提取声纹特征;
    将所述声音波形和预存的标准声音波形进行波形匹配,将所述声纹特征和预存的标准声纹特征进行特征匹配;以及
    如果所述波形匹配和所述特征匹配都匹配成功,则声纹验证通过。
  2. 如权利要求1所述的方法,在所述采集待验证的语音信息之前,所述方法还包括:
    截取用户输入的语音片段;以及
    将所述语音片段的声音波形存储为所述预存的标准声音波形。
  3. 如权利要求2所述的方法,在所述采集待验证的语音信息之前,所述方法还包括:
    根据截取的所述语音片段生成随机密码并存储所述随机密码。
  4. 如权利要求3所述的方法,所述采集待验证的语音信息,包括:
    获取预先存储的随机密码;
    提示用户通过语音方式输入获取的所述随机密码;以及
    采集所述用户通过语音方式输入的所述随机密码,作为待验证的语音信息。
  5. 如权利要求1所述的方法,其中,将所述声音波形和预存的标准声音波形进行波形匹配,将所述声纹特征和预存的标准声纹特征进行特征匹配,包括:
    将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配不成功,则声纹验证不通过;如果所述波形匹配成功,则将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配成功,则声纹验证通过, 如果所述特征匹配不成功,则声纹验证不通过;
    或者,
    将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配不成功,则声纹验证不通过;如果所述特征匹配成功,则将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配成功,则声纹验证通过,如果所述波形匹配不成功,则声纹验证不通过。
  6. 如权利要求1所述的方法,其中,在所述语音信息对应的声音波形中,提取声纹特征,包括:
    将所述声音波形转换为声波频谱,在所述声波频谱中提取声纹特征。
  7. 一种声纹验证装置,包括:
    采集模块,设置为采集待验证的语音信息;
    提取模块,设置为在所述语音信息对应的声音波形中,提取声纹特征;以及
    验证模块,设置为将所述声音波形和预存的标准声音波形进行波形匹配,将所述声纹特征和预存的标准声纹特征进行特征匹配;以及如果所述波形匹配和所述特征匹配都匹配成功,则声纹验证通过。
  8. 如权利要求7所述的装置,其中,所述采集模块,还设置为:
    在采集待验证的语音信息之前,截取用户输入的语音片段;以及
    将所述语音片段的声音波形存储为所述预存的标准声音波形。
  9. 如权利要求8所述的装置,其中,所述采集模块,还设置为:在所述采集待验证的语音信息之前,根据截取的所述语音片段生成随机密码并存储所述随机密码。
  10. 如权利要求9所述的装置,其中,所述采集模块设置为:
    获取预先存储的随机密码;
    提示用户通过语音方式输入获取的所述随机密码;以及
    采集所述用户通过语音方式输入的所述随机密码,作为待验证的语音信息。
  11. 如权利要求7所述的装置,其中,所述验证模块设置为:
    将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配不成功,则声纹验证不通过;如果所述波形匹配成功,则将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配成功,则声纹验证通过,如果所述特征匹配不成功,则声纹验证不通过;
    或者,
    将所述声纹特征和预存的标准声纹特征进行特征匹配,如果所述特征匹配不成功,则声纹验证不通过;如果所述特征匹配成功,则将所述声音波形和预存的标准声音波形进行波形匹配,如果所述波形匹配成功,则声纹验证通过,如果所述波形匹配不成功,则声纹验证不通过。
  12. 如权利要求7所述的方法,其中,在所述语音信息对应的声音波形中,提取声纹特征,包括:
    将所述声音波形转换为声波频谱,在所述声波频谱中提取声纹特征。
  13. 一种计算机可读存储介质,存储有计算机可执行指令,所述计算机可执行指令设置为执行权利要求1-6中任一项的方法。
PCT/CN2017/090171 2017-01-11 2017-06-27 声纹验证方法和装置 WO2018129869A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710019325.8 2017-01-11
CN201710019325.8A CN108319829B (zh) 2017-01-11 2017-01-11 一种声纹验证方法和装置

Publications (1)

Publication Number Publication Date
WO2018129869A1 true WO2018129869A1 (zh) 2018-07-19

Family

ID=62839101

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/090171 WO2018129869A1 (zh) 2017-01-11 2017-06-27 声纹验证方法和装置

Country Status (2)

Country Link
CN (1) CN108319829B (zh)
WO (1) WO2018129869A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111371755A (zh) * 2020-02-24 2020-07-03 平安科技(深圳)有限公司 声纹数据处理方法、装置、计算机设备和存储介质
CN113051537A (zh) * 2021-03-23 2021-06-29 深圳市声扬科技有限公司 身份验证方法、装置、电子设备及计算机可读介质
CN114826709A (zh) * 2022-04-15 2022-07-29 马上消费金融股份有限公司 身份认证和声学环境检测方法、系统、电子设备及介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109344587A (zh) * 2018-08-09 2019-02-15 平安科技(深圳)有限公司 案件处理意见录入方法、装置、计算机设备及存储介质
CN109147797B (zh) * 2018-10-18 2024-05-07 平安科技(深圳)有限公司 基于声纹识别的客服方法、装置、计算机设备及存储介质
CN112580007B (zh) * 2019-09-29 2023-01-10 北京安云世纪科技有限公司 一种验证方法及装置
CN111785280A (zh) * 2020-06-10 2020-10-16 北京三快在线科技有限公司 身份认证方法和装置、存储介质和电子设备
CN112530441A (zh) * 2020-10-23 2021-03-19 深圳市声扬科技有限公司 合法用户的验证方法、装置、计算机设备和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100100376A1 (en) * 2008-10-17 2010-04-22 International Business Machines Corporation Visualization interface of continuous waveform multi-speaker identification
CN102110438A (zh) * 2010-12-15 2011-06-29 方正国际软件有限公司 一种基于语音的身份认证方法及系统
CN102402985A (zh) * 2010-09-14 2012-04-04 盛乐信息技术(上海)有限公司 提高声纹识别安全性的声纹认证系统及其实现方法
CN104217152A (zh) * 2014-09-23 2014-12-17 陈包容 一种移动终端在待机状态下进入应用程序的实现方法和装置
CN105718781A (zh) * 2016-02-23 2016-06-29 惠州Tcl移动通信有限公司 基于声纹识别的操作终端设备的方法及终端设备

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1164576B1 (en) * 2000-06-15 2006-02-22 Swisscom Fixnet AG Speaker authentication method and system from speech models
JP3678417B2 (ja) * 2002-04-26 2005-08-03 正幸 糸井 個人認証方法及びシステム
EP2433246A4 (en) * 2009-05-18 2013-07-03 Mikoh Corp BIOMETRIC IDENTIFICATION PROCEDURES
CN101794386B (zh) * 2010-03-24 2012-10-24 成都方程式电子有限公司 抵御指纹残留的指纹识别系统和方法
US9118669B2 (en) * 2010-09-30 2015-08-25 Alcatel Lucent Method and apparatus for voice signature authentication
US9444816B2 (en) * 2011-03-30 2016-09-13 Qualcomm Incorporated Continuous voice authentication for a mobile device
CN103077713B (zh) * 2012-12-25 2019-02-01 青岛海信电器股份有限公司 一种语音处理方法及装置
CN104036780B (zh) * 2013-03-05 2017-05-24 阿里巴巴集团控股有限公司 一种人机识别方法及系统
CN104092653B (zh) * 2014-01-20 2017-01-25 腾讯科技(深圳)有限公司 一种数据处理的方法和系统
CN103886236A (zh) * 2014-03-17 2014-06-25 深圳市中兴移动通信有限公司 声控解锁屏方法和移动终端
CN104978507B (zh) * 2014-04-14 2019-02-01 中国石油化工集团公司 一种基于声纹识别的智能测井评价专家系统身份认证方法
US10956907B2 (en) * 2014-07-10 2021-03-23 Datalogic Usa, Inc. Authorization of transactions based on automated validation of customer speech
CN104239768B (zh) * 2014-09-04 2018-08-24 深圳市浩方电子商务有限公司 基于生物特征信息验证的个人账户信息安全管理系统及方法
CN105868679B (zh) * 2015-01-23 2017-11-28 深圳市汇顶科技股份有限公司 指纹信息的动态更新方法和指纹识别装置
CN104835498B (zh) * 2015-05-25 2018-12-18 重庆大学 基于多类型组合特征参数的声纹识别方法
CN105096121B (zh) * 2015-06-25 2017-07-25 百度在线网络技术(北京)有限公司 声纹认证方法和装置
CN105141427B (zh) * 2015-08-18 2018-09-14 广州密码科技有限公司 一种基于声纹识别的登录认证方法、装置及系统
CN105701457B (zh) * 2016-01-08 2019-09-24 西安工程大学 基于人脸识别控制的直流电磁继电器装置及其控制方法
CN105791931A (zh) * 2016-02-26 2016-07-20 深圳Tcl数字技术有限公司 智能电视及其语音控制方法
CN105812573A (zh) * 2016-04-28 2016-07-27 努比亚技术有限公司 一种语音处理方法及移动终端
CN105788597A (zh) * 2016-05-12 2016-07-20 深圳市联谛信息无障碍有限责任公司 基于声音识别的读屏应用指令输入方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100100376A1 (en) * 2008-10-17 2010-04-22 International Business Machines Corporation Visualization interface of continuous waveform multi-speaker identification
CN102402985A (zh) * 2010-09-14 2012-04-04 盛乐信息技术(上海)有限公司 提高声纹识别安全性的声纹认证系统及其实现方法
CN102110438A (zh) * 2010-12-15 2011-06-29 方正国际软件有限公司 一种基于语音的身份认证方法及系统
CN104217152A (zh) * 2014-09-23 2014-12-17 陈包容 一种移动终端在待机状态下进入应用程序的实现方法和装置
CN105718781A (zh) * 2016-02-23 2016-06-29 惠州Tcl移动通信有限公司 基于声纹识别的操作终端设备的方法及终端设备

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111371755A (zh) * 2020-02-24 2020-07-03 平安科技(深圳)有限公司 声纹数据处理方法、装置、计算机设备和存储介质
CN111371755B (zh) * 2020-02-24 2023-01-10 平安科技(深圳)有限公司 声纹数据处理方法、装置、计算机设备和存储介质
CN113051537A (zh) * 2021-03-23 2021-06-29 深圳市声扬科技有限公司 身份验证方法、装置、电子设备及计算机可读介质
CN114826709A (zh) * 2022-04-15 2022-07-29 马上消费金融股份有限公司 身份认证和声学环境检测方法、系统、电子设备及介质

Also Published As

Publication number Publication date
CN108319829A (zh) 2018-07-24
CN108319829B (zh) 2022-05-06

Similar Documents

Publication Publication Date Title
WO2018129869A1 (zh) 声纹验证方法和装置
CN108039176B (zh) 一种防录音攻击的声纹认证方法、装置及门禁系统
WO2017114307A1 (zh) 能够防止录音攻击的声纹认证方法、服务器、终端及系统
US10593334B2 (en) Method and apparatus for generating voiceprint information comprised of reference pieces each used for authentication
WO2017197953A1 (zh) 基于声纹的身份识别方法及装置
US20190013026A1 (en) System and method for efficient liveness detection
CN104509065B (zh) 使用说话能力作为人类交互证明
WO2020077885A1 (zh) 身份验证方法、装置、计算机设备和存储介质
US20190279645A1 (en) Reverberation compensation for far-field speaker recognition
CN106537493A (zh) 语音识别系统及方法、客户端设备及云端服务器
US9646613B2 (en) Methods and systems for splitting a digital signal
WO2010047816A1 (en) Speaker verification methods and apparatus
WO2010047817A1 (en) Speaker verification methods and systems
EP3989217B1 (en) Method for detecting an audio adversarial attack with respect to a voice input processed by an automatic speech recognition system, corresponding device, computer program product and computer-readable carrier medium
Wu et al. Vulnerability evaluation of speaker verification under voice conversion spoofing: the effect of text constraints
WO2017215186A1 (zh) 一种安全登录方法和装置、存储介质
CN112382300A (zh) 声纹鉴定方法、模型训练方法、装置、设备及存储介质
CN111091837A (zh) 一种基于在线学习的时变声纹认证方法及系统
CN110379433A (zh) 身份验证的方法、装置、计算机设备及存储介质
Firc et al. The dawn of a text-dependent society: deepfakes as a threat to speech verification systems
Shirvanian et al. Quantifying the breakability of voice assistants
CN112397072B (zh) 语音检测方法、装置、电子设备及存储介质
JP2015079152A (ja) 音声認証装置、音声認証方法及び音声認証プログラム
CN109273012A (zh) 一种基于说话人识别和数字语音识别的身份认证方法
Trysnyuk et al. A method for user authenticating to critical infrastructure objects based on voice message identification

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17891893

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17891893

Country of ref document: EP

Kind code of ref document: A1