Speaker Verification in Realistic Noisy Environment in Forensic Science

Kamada, Toshiaki; Minematsu, Nobuaki; Osanai, Takashi; Makinae, Hisanori; Tanimoto, Masumi

doi:10.1093/ietisy/e91-d.3.558

Cited by 4 publications

(4 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, it is generally considered that speech samples in forensic investigations often contain noise at either A or X, but rarely at both positions. This is because the audio analyzed in forensic science, such as that recorded during a crime, cannot be re-recorded and may contain noise, whereas comparative materials can be recorded later in a clean environment [10].…”

Section: Methodsmentioning

confidence: 99%

“…We created stimulus sounds by superimposing utterances and one of the two types of noise. When superimposing the noise on the utterances, the SNR is calculated based on the root-mean-square (RMS) value [10]. The SNR was either ∞, 0 dB, or −10 dB; for SNR = ∞, we only performed normalization (described below).…”

Section: Experiments 21 Stimulus Sounds Used In Experimentsmentioning

confidence: 99%

See 1 more Smart Citation

Error tendencies of human speaker recognition in noisy environments: Effects of noise type and sex differences

Suzuki,

Amino,

Arai

2024

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

Human speaker recognition performance can be degraded by various factors. Understanding the factors affecting it and the errors caused by these factors is crucial for forensic applications. To study the effects of noisy environments on human speaker recognition, we conducted a hearing experiment using speech samples of two words by five male speakers, and two noise types (speech-like noise and environmental noise in boiler room) with three steps of signal-to-noise ratio (∞, 0 dB, or −10 dB). The results suggested that the listeners tended to observe different speakers to be the same speaker rather than vice versa, and this tendeny was also affected by sex of the listener.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Experiments 21 Stimulus Sounds Used In Experimentsmentioning

confidence: 99%

Error tendencies of human speaker recognition in noisy environments: Effects of noise type and sex differences

Suzuki,

Amino,

Arai

2024

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

show abstract

“…Missing feature theory and spectral subtraction were used in many speaker recognition task in noisy environments [1]- [4]. In [2], a method that combined multicondition model training and missingfeature theory to model noise with temporal-spectral characteristics was proposed.…”

Section: Introductionmentioning

confidence: 99%

“…Some noise-robust techniques, such as noisy speech training models, spectral subtraction and a special missing feature theory are introduced in Sect. 4. The experiments for speaker identification using phase information in noisy conditions are evaluated in Sect.…”

Section: Introductionmentioning

confidence: 99%

Speaker Recognition by Combining MFCC and Phase Information in Noisy Conditions

Wang

Minami

Yamamoto

et al. 2010

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYIn this paper, we investigate the effectiveness of phase for speaker recognition in noisy conditions and combine the phase information with mel-frequency cepstral coefficients (MFCCs). To date, almost speaker recognition methods are based on MFCCs even in noisy conditions. For MFCCs which dominantly capture vocal tract information, only the magnitude of the Fourier Transform of time-domain speech frames is used and phase information has been ignored. High complement of the phase information and MFCCs is expected because the phase information includes rich voice source information. Furthermore, some researches have reported that phase based feature was robust to noise. In our previous study, a phase information extraction method that normalizes the change variation in the phase depending on the clipping position of the input speech was proposed, and the performance of the combination of the phase information and MFCCs was remarkably better than that of MFCCs. In this paper, we evaluate the robustness of the proposed phase information for speaker identification in noisy conditions. Spectral subtraction, a method skipping frames with low energy/Signal-to-Noise (SN) and noisy speech training models are used to analyze the effect of the phase information and MFCCs in noisy conditions. The NTT database and the JNAS (Japanese Newspaper Article Sentences) database added with stationary/non-stationary noise were used to evaluate our proposed method. MFCCs outperformed the phase information for clean speech. On the other hand, the degradation of the phase information was significantly smaller than that of MFCCs for noisy speech. The individual result of the phase information was even better than that of MFCCs in many cases by clean speech training models. By deleting unreliable frames (frames having low energy/SN), the speaker identification performance was improved significantly. By integrating the phase information with MFCCs, the speaker identification error reduction rate was about 30%-60% compared with the standard MFCCbased method. key words: speaker identification, phase information, MFCC, noisy environment, GMM

show abstract

Voice authentication by text dependent single utterance for in-car environment

Tamoto

Itou

2019

Proceedings of the Tenth International Symposium on Information and Communication Technology - SoICT 2019

View full text Add to dashboard Cite

Speaker Verification in Realistic Noisy Environment in Forensic Science

Cited by 4 publications

References 5 publications

Error tendencies of human speaker recognition in noisy environments: Effects of noise type and sex differences

Error tendencies of human speaker recognition in noisy environments: Effects of noise type and sex differences

Speaker Recognition by Combining MFCC and Phase Information in Noisy Conditions

Voice authentication by text dependent single utterance for in-car environment

Contact Info

Product

Resources

About