ID 原文 译文
39226 在免提通话系统和移动通信设备中，扬声器常常工作在较高的音量下，容易发生过载现象，从而产生明显的非线性声学回声，这在小微型扬声器中更加常见。 In hand-free telephone systems and mobile communication devices, the loudspeakers often work with large volume and are easily in overload status, so obvious nonlinear acoustic echo occurs. This phenomenon is more usual in devices with small-or micro-loudspeakers.
39227 常用的线性AEC(Acoustic Echo Cancellation)算法无法消除此类非线性回声，因此通话质量受到严重影响。 The general AEC(Acoustic Echo Cancellation) algorithm cannot cancel the nonlinear echo, thus the communication speech quality degrades significantly.
39228 非线性回声主要表现为额外的高频谐波分量，这些分量使得全带系统不再满足线性关系，而通常的AEC算法都是基于最小化全带误差推导而来，因此性能很容易受到非线性失真的影响。 The nonlinear echo has many extra high-frequency harmonic distortions, and these distortions break the linear constraint of a full-band adaptive filter algorithm.Since the general AEC algorithms are derived to minimize the full-band error, their performances are very sensitive to the nonlinear distortion.
39229 本文提出了一种基于多相滤波器组的子带AEC算法，把全带误差变成了各个子带的误差，因而把谐波失真成分变成了某些子带内的加性噪声，这使得谐波失真较小的那些子带依然能够正常收敛。 This paper proposes a subband AEC algorithms based on polyphase filter-bank, which transforms the fullband error into subband errors, and those harmonic distortions become additive noise in corresponding subbands. In those bands that distortions are small, the subband adaptive filter can convergence normally.
39230 通过仿真和实测实验，当出现非线性失真时，新方法的ERLE(Echo Return Loss Enhancement)明显高于经典的全带时域和频域方法，对于非线性失真明显的语音信号，ERLE提升约10 dB。 In simulation and real tests, the results show that when the nonlinear distortion occurs, the proposed method achieves significantly better ERLE(Echo Return Loss Enhancement) than the typical time-domain and frequency-domain methods. For speech signals with obvious nonlinear distortion, ERLE improves about 10 dB.
39231 音质(Timbre)是音乐感知和言语识别的重要线索。 Timbre is an important clue for music perception and speech recognition.
39232 传统音质分析方法无法同时获取理想的时间分辨率和频域分辨率，对音频的非平稳特性没有很好地处理。 The traditional feature extraction method cannot obtain the ideal temporal resolution and frequency resolution at the same time, and the non-stationary information of audio is not well explored.
39233 本文采用时变滤波经验模态分解(Time Varying Filtering based EMD，TVF-EMD)方法提取音频的固有模态函数用于希尔伯特变换，并构建了音质的希尔伯特频谱分布特征和希尔伯特轮廓特征。 To solve the above problems, the time varying filtering based EMD(TVF-EMD) method was adopted in this paper to extract the intrinsic mode function of audio for the Hilbert Transform, and constructed the Hilbert spectrum distribution features and Hilbert contour features.
39234 在乐器分类问题中，将提取的两类音质特征与Mel倒谱系数特征(Mel Frequency Cepstral Coefficients， MFCCs)有效结合，然后构造基于双向长短时记忆网络的音质时序分类器，在公开乐器演奏音频数据库中进行了乐器分类实验。 In the experiment of musical instrument classification, we combined the two kinds of features with the Mel frequency cepstral coefficients(MFCCs), and then constructed a time sequence classifier based on Bi-directional Long Short-Term Memory(BiLSTM). The experiment of musical instrument classification was carried out in the public musical instrument audio database.
39235 结果表明，所提出的音质特征可以有效补充Mel倒谱特征等传统特征无法表达的非线性非平稳信息，大大提高了本音质表征方法对复杂音频的适应性和鲁棒性。 The experimental results show that the proposed features can supplement the non-linear non-stationary information which is not extracted from the traditional features such as MFCCs, and improve the adaptability and robustness of timbre features to complex audio.