Search

This paper describes a hands-on comparison on using state-of-the-art music source separation deep neural networks (DNNs) before and after task-specific fine-tuning for separating speech content from non-speech content in broadcast audio (i.e., dialog separation). The music separation models are selected as they share the number of channels (2) and sampling rate (44.1 kHz or higher) with the considered broadcast content, and vocals separation in music is considered as a parallel for dialog separation in the target application domain. These similarities are assumed to enable transfer learning between the tasks. Three models pre-trained on music (Open-Unmix, Spleeter, and Conv-TasNet) are considered in the experiments, and fine-tuned with real broadcast data. The performance of the models is evaluated before and after fine-tuning with computational evaluation metrics (SI-SIRi, SI-SDRi, 2f-model), as well as with a listening test simulating an application where the non-speech signal is partially attenuated, e.g., for better speech intelligibility. The evaluations include two reference systems specifically developed for dialog separation. The results indicate that pre-trained music source separation models can be used for dialog separation to some degree, and that they benefit from the fine-tuning, reaching a performance close to task-specific solutions., accepted in INTERSPEECH 2021

The possible fields of application for small sensor nodes are tremendous and still growing fast. Concepts like the Internet of Things (IoT), Smart City or Industry 4.0 adopt wireless sensor networks for environmental interaction or metering purposes. As they commonly operate in license-exempt frequency bands, telemetry transmissions of sensors are subject to strong interferences and possible shadowing. Especially in the scope of Low Power Wide Area (LPWA) communications, this scenario results in high computational effort and complexity for the receiver side to perceive the signals of interest. Therefore, this paper investigates means to an adequate segmentation of receive spectra for a partial spectrum exchange between base stations of telemetry-based IoT sensor networks. The distinct interchange of in-phase and quadrature (IQ) data could facilitate stream combining techniques to mask out interferences amongst other approaches. This shall improve decoding rates even under severe operation conditions and simultaneously limit the required data volume. We refer to this approach of a reception network as Edge-RAN (Random Access Network). To cope with the high data rates and still enable a base station collaboration, especially in wirelessly connected receiver mesh networks, different filter bank techniques and block transforms are examined, to divide telemetry spectra into distinct frequency sub-channels. Operational constraints for the spectral decomposition are given and different filter methodologies are introduced. Finally, suitable metrics are established. These metrics shall assess the performance of the presented spectrum segmentation schemes for the purpose of a selective partial interchange between sensor network receivers.

A blind bandwidth extension is presented which improves the perceived quality of 4 kHz speech by artificially extending the speech’s frequency range to 8 kHz. Based on the source-filter model of the human speech production, the speech signal is decomposed into spectral envelope and excitation signal and each of them is extrapolated separately. With this decomposition, good perceptual quality can be achieved while keeping the computational complexity low. The focus of this work is in the generation of an excitation signal with and autoregressive model that calculates a distribution for each audio sample conditioned on previous samples. This is achieved with a deep neural network following the architecture of LPCNet [1].A listening test shows that it significantly improves the perceived quality of bandlimited speech. The system has an algorithmic delay of 30 ms and can be applied in state-of-the-art speech and audio codecs.

In this paper, we investigate the coding efficiency of perceptual coding using an adaptive non-uniform orthogonal filter-bank based on MDCT analysis/synthesis and time domain aliasing reduction. We compare its performance to a system using a traditional adaptive uniform MDCT filterbank with window switching. The comparison is performed using a listening test at two different quantization settings. The statistical evaluation shows that the percetpual quality of the nonuniform filterbank significantly out-performs that of the uniform filterbank by 5 to 10 MUSHRA points.

Perceptual audio coding schemes typically apply the modified discrete cosine transform (MDCT) with different lengths and windows, and utilize signal-adaptive switching between these on a perframe basis for best subjective performance. In previous papers, the authors demonstrated that further quality gains can be achieved for some input signals using additional transform kernels such as the modified discrete sine transform (MDST) or greater inter-transform overlap by means of a modified extended lapped transform (MELT). This work discusses the algorithmic procedures and codec modifications necessary to combine all of the above features—transform length, window shape, transform kernel, and overlap ratio switching—into a flexible input-adaptive coding system. It is shown that, due to full time-domain aliasing cancelation, this system supports perfect signal reconstruction in the absence of quantization and, thanks to fast realizations of all transforms, increases the codec complexity only negligibly. The results of a 5.1 multichannel listening test are also reported.

International audience; Estimating the maximum number of concurrent speakers from single-channel mixtures is a challenging problem and an essential first step to address various audio-based tasks such as blind source separation, speaker diarization, and audio surveillance. We propose a unifying probabilistic paradigm, where deep neural network architectures are used to infer output posterior distributions. These probabilities are in turn processed to yield discrete point estimates. Designing such architectures often involves two important and complementary aspects that we investigate and discuss. First, we study how recent advances in deep architectures may be exploited for the task of speaker count estimation. In particular, we show that convolutional recurrent neural networks outperform recurrent networks used in a previous study when adequate input features are used. Even for short segments of speech mixtures, we can estimate up to five speakers, with a significantly lower error than other methods. Second, through comprehensive evaluation, we compare the best-performing method to several baselines, as well as the influence of gain variations, different data sets, and reverberation. The output of our proposed method is compared to human performance. Finally, we give insights into the strategy used by our proposed method.

Dereverberation is useful in hands-free communication and voice controlled devices for distant speech acquisition. Single-channel dereverberation can be achieved by applying a time-frequency (TF) mask to the short-time Fourier transform (STFT) representation of a reverberant signal. Recent approaches have used deep neural networks (DNNs) to estimate such masks. Previously proposed DNN-based mask estimation methods train a DNN to minimize the mean-squared-error (MSE) between the desired and estimated masks. Recent TF mask estimation methods for signal separation directly minimize instead the MSE between the desired and estimated STFT magnitudes. We apply this direct optimization concept to dereverberation. Moreover, as reverberation exceeds the duration of a single STFT frame, we propose to use a bidirectional long short-term memory (LSTM) network which is able to take the relation between multiple STFT frames into account. We evaluated our method for different reverberation times and source-microphone distances using simulated as well as measured room impulse responses of different rooms. An evaluation of the proposed method and a comparison with a state-of-the-art method demonstrate the superiority of our approach and its robustness to different acoustic conditions.

For a long time, many popular listening test methods, such as ITU-R BS.1534 (MUSHRA), could not be carried out as web-based listening tests, since established web standards did not support all required audio processing features. With the standardization of the Web Audio API, the required features became available and, therefore, also the possibility to implement a wide range of established methods as web-based listening tests. In order to simplify the implementation of MUSHRA listening tests, the development of webMUSHRA was started. By utilizing webMUSHRA, experimenters can configure web-based MUSHRA listening tests without the need of web programming expertise. Today, webMUSHRA supports many more listening test methods, such as ITU-R BS.1116 and forced-choice procedures. Moreover, webMUS HRA is highly customizable and has been used in many auditory studies for different purposes.

The task of estimating the maximum number of concurrent speakers from single channel mixtures is important for various audio-based applications, such as blind source separation, speaker diarisation, audio surveillance or auditory scene classification. Building upon powerful machine learning methodology, we develop a Deep Neural Network (DNN) that estimates a speaker count. While DNNs efficiently map input representations to output targets, it remains unclear how to best handle the network output to infer integer source count estimates, as a discrete count estimate can either be tackled as a regression or a classification problem. In this paper, we investigate this important design decision and also address complementary parameter choices such as the input representation. We evaluate a state-of-the-art DNN audio model based on a Bi-directional Long Short-Term Memory network architecture for speaker count estimations. Through experimental evaluations aimed at identifying the best overall strategy for the task and show results for five seconds speech segments in mixtures of up to ten speakers., Accepted in ICASSP 2018

In this letter we describe nonuniform orthogonal modified discrete cosine transform (MDCT) filterbanks and time-domain aliasing reduction (TDAR). By adding a postprocessing step to the MDCT, our method allows for arbitrary nonuniform frequency resolutions using subband merging with smooth windowing and overlap in frequency. This overlap allows for an improved temporal compactness of the impulse response, which is especially useful for audio coders. The postprocessing step comprises another lapped MDCT transform along the frequency axis and TDAR along each subband signal.

This paper examines how masked thresholds depend on the masker bandwidth and center frequency when the masker has a smaller bandwidth than the signal. The signal bandwidth was equal to the equivalent rectangular bandwidth of the auditory filter and the masker bandwidth was 0.1, 0.35, or 0.6 times the signal bandwidth. The masker and signal were centered at the same frequency of 257, 697, 1538, 3142, or 6930 Hz. Masked thresholds were estimated using a two-interval two-alternative forced-choice paradigm and a three-down one-up adaptive staircase method. Masked thresholds increased with increasing masker bandwidth and were lowest for medium center frequencies.

Advanced Bionics' cochlear implants have independent current sources that can share stimulation current between 2 contacts (Current Steering). By stimulating 2 adjacent electrodes with different weights, different pitches can be evoked, allowing to increase the number of processing channels.A counterbalanced crossover design was used to compare 3 different current steering implementations to the clinical HiRes strategy.The study was a prospective, within-subject, repeated-measure experiment.The study group consisted of 8 postlingually deaf subjects with a minimum of 12-month experience in HiRes.The following programs were evaluated: 1) a Fast Fourier Transformation (FFT)-based current steering implementation with 120 stimulation sites; 2) the same current steering implementation but with 16,000 stimulation sites; and 3) a current steering implementation based on a sinusoidal decomposition of the original signal with 16,000 stimulation sites.Speech perception tests in quiet and in Comité Consultatif International Télégraphique et Téléphonique (CCITT) noise, as well as with competing talker; an adaptive test of the frequency difference limen; a Quality Assessment Questionnaire.Current results do not show any improvement in speech perception for a certain current steering strategy compared with HiRes. However, when selecting the optimal current steering strategy, subjects could achieve a significant benefit compared with the clinical HiRes. In addition, the frequency difference limen could be reduced significantly at 1,280 Hz.Current steering seems to have the potential to improve both understanding in adverse listening situations and frequency resolution. However, the optimal implementation needs further investigation.

International audience; In this paper we present a novel source separation method aiming to overcome the difficulty of modelling non-stationary signals. The method can be applied to mixtures of musical instruments with frequency and/or amplitude modulation, e.g. typically caused by vi-brato. It is based on a signal representation that divides the complex spectrogram into a grid of patches of arbitrary size. These complex patches are then processed by a two-dimensional discrete Fourier transform, forming a tensor representation which reveals spectral and temporal modulation textures. Our representation can be seen as an alternative to modulation transforms computed on magnitude spectrograms. An adapted factorization model allows to decompose different time-varying harmonic sources based on their particular common modulation profile: hence the name Common Fate Model. The method is evaluated on musical instrument mixtures playing the same fundamental frequency (unison), showing improvement over other state-of-the-art methods.

Perceptual audio coders exploit the masking properties of the human auditory system to reduce the bit rate in audio recording and transmission systems; it is intended that the quantization noise is just masked by the audio signal. The effectiveness of the audio signal as a masker depends on whether it is tone-like or noise-like. The determination of this, both physically and perceptually, depends on the duration of the stimuli. To gather information that might improve the efficiency of perceptual coders, the duration required to distinguish between a narrowband noise and a tone was measured as a function of center frequency and noise bandwidth. In experiment 1, duration thresholds were measured for isolated noise and tone bursts. In experiment 2, duration thresholds were measured for tone and noise segments embedded within longer tone pulses. In both experiments, center frequencies were 345, 754, 1456, and 2658 Hz and bandwidths were 0.25, 0.5, and 1 times the equivalent rectangular bandwidth of the auditory filter at each center frequency. The duration thresholds decreased with increasing bandwidth and with increasing center frequency up to 1456 Hz. It is argued that the duration thresholds depended mainly on the detection of amplitude fluctuations in the noise bursts.

Modern stereo and multi-channel perceptual audio codecs utilizing the modified discrete cosine transform (MDCT) can achieve very good overall coding quality even at low bit-rates but lack efficiency on some material with inter-channel phase difference (IPD) of about ±90 degrees. To address this issue a generalization of the lapped transform coding scheme is proposed which retains the perfect reconstruction property while allowing the usage of three further transform kernels, one of which is the modified discrete sine transform (MDST). Blind listening tests indicate that by frame-wise adaptation of each channel's transform kernel to the instantaneous IPD characteristics, notable gains in coding quality are possible with only negligible increase in decoder complexity and parameter rate.

Traditional audio codecs based on real-valued transforms utilize separate and largely independent algorithmic schemes for parametric coding of noise-like or high-frequency spectral components as well as channel pairs. It is shown that in the frequency-domain part of coders such as Extended HE-AAC, these schemes can be unified into a single algorithmic block located at the core of the modified discrete cosine transform path, enabling greater flexibility like semi-parametric coding and large savings in codec delay and complexity. This paper focuses on the stereo coding aspect of this block and demonstrates that, by using specially chosen spectral configurations when deriving the parametric side-information in the encoder, perceptual artifacts can be reduced and the spatial processing in the decoder can remain real-valued. Listening tests confirm the benefit of our proposal at intermediate bit-rates.

Perceptual audio codecs apply psychoacoustic principles such as masking effects of the human auditory system in order to reduce irrelevancies in the input audio signal. Psychoacoustic studies show differences between masking strength of tonal and noise maskers: the masking effect of narrowband noise is stronger than that of a tone which has the same power and is placed in the center frequency of the noise. In this paper, two tonality estimation methods are discussed which are implemented in a filter bank based psychoacoustic model. The first method is called Partial Spectral Flatness Measure (PSFM) and the second is referred to as Amplitude Modulation Ratio (AM-R). The psychoacoustic model uses a set of complex band-pass filters. It was designed according to the temporal/spectral resolution of the human auditory system, and takes into account post masking as well as the spreading effect of individual local maskers in simultaneous masking. This paper describes the model, tonality estimation methods and their implementation. The estimators are compared to each other by subjective tests. The results are presented and discussed.

General-purpose MDCT-based audio coders like MP3 or HE-AAC utilize long inter-transform overlap and lookahead-based transform length switching to provide good coding quality for both stationary and non-stationary, i. e. transient, input signals even at low bitrates. In low-delay communication scenarios such as Voice over IP, however, algorithmic delay due to framing and overlap typically needs to be reduced and additional lookahead must be avoided. We show that these restrictions limit the performance of contemporary low-delay transform coders on either stationary or transient material and propose 3 modifications: an improved noise substitution technique and increased overlap between “long”transforms for stationary, and “long to short” transform length switching without lookahead and directly from the long overlap for transient frames. A listening test indicates the merit of these changes when integrated into AAC-LD.

Modern transform audio coders often employ parametric enhancements, like noise substitution or bandwidth extension. In addition to these well-known parametric tools, it might also be desirable to synthesize parametric sinusoidal tones in the decoder. Low computational complexity is an important criterion in codec development and essential for acceptance and deployment. Therefore, efficient ways of generating these tones are needed. Since contemporary codecs like AAC or USAC are based on an MDCT domain representation of audio, we propose to generate synthetic tones by patching tone patterns into the MDCT spectrum at the decoder. We demonstrate how appropriate spectral patterns can be derived and adapted to their target location in (and between) the MDCT time/frequency (t/f) grid to seamlessly synthesize high quality sinusoidal tones including sweeps.

Psychoacoustic studies show that a narrowband noise masker exhibits a stronger simultaneous masking effect than a tonal masker with the same signal power placed at the noise center frequency. Consequently, perceptual audio codecs commonly incorporate some sort of tonality estimation as part of their perceptual model. However, common tonality estimation techniques do not necessarily reflect the perception of tonality by human listeners. As long as the tone and narrowband noise signals are long enough, they are easily distinguishable for normal hearing listeners. However, if the stimulus duration decreases, both signal types approach the shape of impulses and therefore, at some point become audibly identical. Consequently, at a given frequency and noise bandwidth, there is a duration threshold below which the signals cannot be distinguished. A series of so-called "2-AFC 3-step up-down" psychoacoustic tests are designed and carried out to investigate the frequency and bandwidth dependency of these duration thresholds. The test results, collected from 32 listeners, are statistically evaluated and confirm a decreasing threshold for increasing center frequency and bandwidth. These results can be used to improve psychoacoustic models for audio codecs by using tonality estimators with frequency and bandwidth adapted temporal resolution.

Modern music production often uses pre-recorded pieces of audio, so-called samples, taken from a huge sample database. Consequently, there is an increasing demand to extensively adapt these samples to their intended new musical environment in a flexible way. Such an application, for instance, retroactively changes the key mode of audio recordings, e.g. from a major key to minor key by a frequency selective transposition of pitch. Recently, the modulation vocoder (MODVOC) has been proposed to handle this task. In this paper, two enhancements to the MODVOC are presented and the subjective quality of its application to selective pitch transposition is assessed. Moreover, the proposed scheme is compared with results obtained by applying a commercial computer program, which became newly available on the market. The proposed method is clearly preferred in terms of the perceptual quality aspect ≪melody and chords transposition≫, while the commercial program is favored by the majority with regard to the aspect ≪timbre preservation≫.

The decomposition of audio signals into perceptually meaningful multiband modulation components opens up new possibilities for advanced signal processing. The signal adaptive analysis approach proposed in this paper will be shown to provide a powerful handle on the signal's perceptual properties: pitch, timbre or roughness can be manipulated straight forward. Additionally a synthesis method is specified providing high subjective perceptual quality. Furthermore, as an application example, a novel audio processing technique is proposed which changes the key mode of a given piece of music e.g. from major to minor key or vice versa.

OBJECTIVE In normal-hearing listeners acoustic masking occurs depending on frequency, amplitude, and energy of specific signals. If the selection of stimulated channels in cochlear implant systems was based on psychoacoustic masking models, the bandwidth of the electrode/nerve interface could be used more effectively by concentrating on relevant signal components and neglecting those that are usually not perceived by normal hearing listeners. Subsequently, a new strategy called PACE (Psychoacoustic Advanced Combination Encoder) has been developed which uses a psychoacoustic model for the channel selection instead of the simple maxima selection algorithm of the ACE strategy. STUDY DESIGN Only subjects having at least 2 years experience with the ACE strategy were included. A counterbalanced cross-over design was used to compare the new speech coding strategy with the ACE strategy. SETTING The investigation was a prospective, within-subject, repeated-measures experiment. PATIENTS The study group consisted of 10 postlingually deafened adult subjects. INTERVENTIONS The following programs were evaluated: (1) ACE with 8 maxima selected; (2) PACE with 8 channels selected; and (3) PACE with 4 channels selected. MAIN OUTCOME MEASURES Speech perception tests in quiet and noise, Quality Assessment Questionnaire. RESULTS Results indicate a trend towards better performance with PACE. Scores in the Freiburg monosyllabic word test increased by 8% while the SNR50 in the Oldenburger sentence test improved significantly by 1.3 dB. CONCLUSION The use of psychoacoustic masking models in speech coding strategies has the potential to improve speech perception performance in cochlear implant subjects.

Efficient combinations of coding and manipulation of audio signals in the spectral domain are often desirable in communication systems. The modified discrete cosine transform (MDCT) represents a popular spectral transform in audio coding as it leads to compact signal representations. However, as the MDCT corresponds to a critically sampled filter bank, it is in general not appropriate to directly apply it to filtering tasks. In this paper we present a method to compensate for aliasing terms that arise from such direct MDCT domain filtering. The discussion is thereby based on a rigorous matrix representation of critically sampled filter banks which also leads to corresponding efficient realizations. As an application showcase, noise reduction for MDCT based speech coding is considered in simulations.

Today, cochlear implants (CIs) are the treatment of choice in patients with profound hearing loss. However speech intelligibility with these devices is still limited. A factor that determines hearing performance is the processing method used in CIs. Therefore, research is focused on designing different speech processing methods. The evaluation of these strategies is subject to variability as it is usually performed with cochlear implant recipients. Hence, an objective method for the evaluation would give more robustness compared to the tests performed with CI patients. This paper proposes a method to evaluate signal processing strategies for CIs based on a hidden markov model speech recognizer. Two signal processing strategies for CIs, the Advanced Combinational Encoder (ACE) and the Psychoacoustic Advanced Combinational Encoder (PACE), have been compared in a phoneme recognition task. Results show that PACE obtained higher recognition scores than ACE as found with CI recipients.

A physiological and computational model of the human auditory system has been fitted in a signal processing strategy for cochlear implants (CIs). The aim of the new strategy is to obtain more natural sound in CIs by better mimicking the human auditory system. The new strategy was built in three independent stages as proposed in [6]. First a basilar membrane motion model was substituted by the filterbank commonly used in commercial strategies. Second, an inner hair cell model was included in a commercial strategy while maintaining the original filterbank. Third, both the basilar membrane motion and the inner-hair cell model were included in the commercial strategy. This paper analyses the properties and presents results obtained with CI recipients for each algorithm designed.

Current speech processing in cochlear implants use a filterbank to analyse audio signals into several frequency bands, each associated with one electrode. Because the processing is performed on input signal blocks of fixed sizes, the filterbank provides a unique time-frequency resolution to represent the various signal features. However, different components of audio signals may require different time-frequency resolutions for an accurate representation and perception. In this paper we investigate the influence on speech intelligibility in cochlear implants users when filterbanks with different time-frequency resolutions are used. In order to represent all signal features accurately, an adaptive filterbank has been developed that accepts input blocks of different sizes. The different resolutions required are achieved by adequately switching between block sizes depending on the input signal characteristics. The filterbank was incorporated into the commercial Advanced Combinational Encoder (ACE) and acutely tested on six cochlear implant recipients.

Current speech processing strategies for cochlear implants use a filterbank which decomposes the audio signals into multiple frequency bands each associated with one electrode. Pitch perception with cochlear implants is related to the number of electrodes inserted in the cochlea and to the rate of stimulation of these electrodes. The filterbank should, therefore, be able to analyze the time-frequency features of the audio signals while also exploiting the time-frequency features of the implant. This study investigates the influence on speech intelligibility in cochlear implant users when filterbanks with different time-frequency resolutions are used. Three filter-banks, based on the structure of a wavelet packet transform but using different basis functions, were designed. The filter-banks were incorporated into a commercial speech processing strategy and were tested on device users in an acute study.

Standard video compression techniques apply motion-compensated prediction combined with transform coding of the prediction error. In the context of prediction with fractional-pel motion vector resolution it was shown, that aliasing components contained in an image signal are limiting the prediction accuracy obtained by motion compensation. In order to consider aliasing, quantisation and motion estimation errors, camera noise, etc., we analytically developed a two-dimensional (2D) non-separable interpolation filter, which is calculated for each frame independently by minimising the prediction error energy. For every fractional-pel position to be interpolated, an individual set of 2D filter coe±cients is determined. Since transmitting filter coefficients as side information results in an additional bit rate, which is almost independent for different total bit rates and image resolutions, the overall gain decreases when total bit rates decrease. In this paper we present an algorithm, which regards the non-separable two-dimensional filter as a polyphase filter. For each frame, predicting the interpolation filter impulse response through evaluation of the polyphase filter, we only have to encode the filter coefficients prediction error. This enables bit rate savings, needed for transmitting filter coe±cients of up to 75% compared to PCM coding. A coding gain of up to 1,2 dB Y-PSNR at same bit rate or up to 30% reduction of bit rate is obtained for HDTV-sequences compared to the standard H.264/AVC. Up to 0,5 dB (up to 10% bit rate reduction) are achieved for CIF-sequences.

For very low bit rate audio coding applications in mobile communications or on the Internet, parametric audio coding has evolved as a technique complementing the more traditional approaches. These are transform codecs originally designed for achieving CD-like quality on one hand, and specialized speech codecs on the other hand. Both of these techniques usually represent the audio signal waveform in a way such that the decoder output signal gives an approximation of the encoder input signal, while taking into account perceptual criteria. Compared to this approach, in parametric audio coding the models of the signal source and of human perception are extended. The source model is now based on the assumption that the audio signal is the sum of "components," each of which can be approximated by a relatively simple signal model with a small number of parameters. The perception model is based on the assumption that the sound of the decoder output signal should be as similar as possible to that of the encoder input signal. Therefore, the approximation of waveforms is no longer necessary. This approach can lead to a very efficient representation. However, a suitable set of models for signal components, a good decomposition, and a good parameter estimation are all vital for achieving maximum audio quality. We give an overview on the current status of parametric audio coding developments and demonstrate advantages and challenges of this approach. Finally, we indicate possible directions of further improvements.

A novel concept for perceptual audio coding is presented which is based on the combination of a pre- and post-filter, controlled by a psychoacoustic model, with a transform coding scheme. This paradigm allows modeling of the temporal and spectral shape of the masked threshold with a resolution independent of the used transform. By using frequency warping techniques the maximum possible detail for a given filter order can be made frequency-dependent and thus better adapted to the human auditory system. The filter coefficients are represented efficiently by LSF parameters which can be adaptively interpolated over time. First experiments with a system obtained by extending an existing transform codec showed that this approach can significantly improve the performance for speech signals, while the performance for other signals remained the same.

This paper proposes a versatile perceptual audio coding method that achieves high compression ratios and is capable of low encoding/decoding delay. It accommodates a variety of source signals (including both music and speech) with different sampling rates. It is based on separating irrelevance and redundancy reductions into independent functional units. This contrasts traditional audio coding where both are integrated within the same subband decomposition. The separation allows for the independent optimization of the irrelevance and redundancy reduction units. For both reductions, we rely on adaptive filtering and predictive coding as much as possible to minimize the delay. A psycho-acoustically controlled adaptive linear filter is used for the irrelevance reduction, and the redundancy reduction is carried out by a predictive lossless coding scheme, which is termed weighted cascaded least mean squared (WCLMS) method. Experiments are carried out on a database of moderate size which contains mono-signals of different sampling rates and varying nature (music, speech, or mixed). They show that the proposed WCLMS lossless coder outperforms other competing lossless coders in terms of compression ratios and delay, as applied to the pre-filtered signal. Moreover, a subjective listening test of the combined pre-filter/lossless coder and a state-of-the-art perceptual audio coder (PAC) shows that the new method achieves a comparable compression ratio and audio quality with a lower delay.

We describe a new signal processing technique for cochlear implants using a psychoacoustic-masking model. The technique is based on the principle of a so-called "NofM" strategy. These strategies stimulate fewer channels ( ) per cycle than active electrodes (NofM; ). In "NofM" strategies such as ACE or SPEAK, only the channels with higher amplitudes are stimulated. The new strategy is based on the ACE strategy but uses a psychoacoustic-masking model in order to determine the essential components of any given audio signal. This new strategy was tested on device users in an acute study, with either 4 or 8 channels stimulated per cycle. For the first condition (4 channels), the mean improvement over the ACE strategy was . For the second condition (8 channels), no significant difference was found between the two strategies.

Das hier vorgestellte Verfahren der Bitratenreduktion für Audiosignale basiert auf überlappenden Transformationen mit „Time Domain Aliasing Cancellation“, deren Fensterfunktionen und Transformationslängen in Abhängigkeit vom Eingangssignal umgeschaltet werden. Die adaptive Fensterung verbessert das Verhalten der Transformationscodierung mit überlappenden Blöcken, die sich durch einen hohen Codierungsgewinn auszeichnet, beim Auftreten von Impulsen und Amplitudensprüngen im Eingangssignal. © 1989, Walter de Gruyter. Alle Rechte vorbehalten.