Simulated Room Impulse Responses Research Articles

Robust speech recognition in reverberant environments by using an optimal synthetic room impulse response model

This paper presents a practical technique for Automatic speech recognition (ASR) in multiple reverberant environment selection. Multiple ASR models are trained with artificial synthetic room impulse responses (IRs), i.e. simulated room IRs, with different reverberation time (T60Models) and tested on real room IRs with varying T60Rooms. To apply our method, the biggest challenge is to choose a proper artificial room IR model for training ASR models. In this paper, a generalised statistical IR model with attenuated reverberation after an early reflection period, named attenuated IR model, has been adopted based on three time-domain statistical IR models. Its optimal values of the reverberation-attenuation factor and the early reflection period on the recognition rate have been searched and determined. Extensive testing has been performed over four real room IR sets (63 IRs in total) with variant T60Rooms and speaker microphone distances (SMDs). The optimised attenuated IR model had the best performance in terms of recognition rate over others. Specific considerations of the practical use of the method have been taken into account including: (i) the maximal training step of T60Model in order to get the minimal number of models with acceptable performance; (ii) the impact of selection errors on the ASR caused by the estimation error of T60Room; and (iii) the performance over SMD and direct-to-reverberation energy Ratio (DRR). It is shown that recognition rates of over 80∼90% are achieved in most cases. One important advantage of the method is that T60Room can be estimated either from reverberant sound directly (Takeda et al., 2009; Falk and Chan, 2010; Löllmann et al., 2010) or from an IR measured from any point of the room as it remains constant in the same room (Kuttruff, 2000), thus it is particularly suited to mobile applications. Compared to many classical dereverberation methods, the proposed method is more suited to ASR tasks in multiple reverberant environments, such as human-robot interaction.

Simulated acoustical environments for the evaluation of vocal effort

Realistic simulations of acoustical environments allow researchers to quickly manipulate the auditory experiences of human subjects. For studies investigating human speakers’ perceptions of their own voices, a mixed-reality environment is most suitable. In this work, we have developed a system to create low-latency real-time convolutions of speakers’ voices with simulated room impulse responses at their ears. The latter are based in part on measured voice directivity patterns. The simulated rooms included classrooms, lecture halls, and auditoria. They were all simulated for human speakers within an anechoic chamber. We also added realistic noise to the simulations, including chatter and other ambient effects, and measured the subjects’ vocal efforts. The ultimate aim of the simulation and measurement system is to assess teachers’ vocal efforts in classrooms and other settings with easily controlled acoustical conditions. However, the setup may also be easily adapted to other studies for speech or music.

Combined wave and ray based room acoustic simulations of audio systems in car passenger compartments, Part I: Boundary and source data

The present series of papers summarizes the results of a three-year research project on the realistic simulation of car audio sound in car passenger compartments using a combined Finite Element (FE) and Geometrical Acoustics (GA) approach. The simulations are conducted for the whole audible frequency range with the loudspeakers of the car audio system as the sound sources. The challenges faced during the project relate to fundamental questions regarding the realistic sound field simulation in small enclosures with strong modal and diffraction effects.The paper denoted here as Part I focuses on boundary and source representations in the FE and GA domain and suggests guidelines for a best-possible acquisition of the required data. Since a straight-forward determination of the boundary and source characteristics is mostly hampered by the immense complexity and inhomogeneity of the materials and loudspeaker configurations inside a car compartment, different measurement and calculation methods have been applied to determine the required data and quantify the corresponding uncertainty. The paper clearly points out the strength and weaknesses of the applied methods depending on the considered frequency range and material characteristics. In order to keep the complexity of the FE simulations at a manageable level, all passive boundaries were considered as locally reacting with impedance conditions.Part II of the study applies the obtained data in combined FE-GA room acoustic simulations and compares the simulated room impulse responses (RIR) with corresponding measurement results. In a final step the observed differences in the RIRs are related to the uncertainty and inherent errors in the boundary and source representation.

Rigid sphere room impulse response simulation: Algorithm and applications

Simulated room impulse responses have been proven to be both useful and indispensable for comprehensive testing of acoustic signal processing algorithms while controlling parameters such as the reverberation time, room dimensions, and source-array distance. In this work, a method is proposed for simulating the room impulse responses between a sound source and the microphones positioned on a spherical array. The method takes into account specular reflections of the source by employing the well-known image method, and scattering from the rigid sphere by employing spherical harmonic decomposition. Pseudocode for the proposed method is provided, taking into account various optimizations to reduce the computational complexity. The magnitude and phase errors that result from the finite order spherical harmonic decomposition are analyzed and general guidelines for the order selection are provided. Three examples are presented: an analysis of a diffuse reverberant sound field, a study of binaural cues in the presence of reverberation, and an illustration of the algorithm's use as a mouth simulator.

Estimating the direct-to-reverberant energy ratio from the coherence between coincident pressure and particle velocity

An analytical expression for the relationship between the direct-to-reverberant energy ratio (DRR) and the coherence estimation function between coincident pressure and particle velocity component is derived. The analytical solution is first validated with simulated room impulse responses and then used to estimate the DRR in five octave bands for several receiver positions measured in a total of 11 rooms of vastly different sizes and acoustic characteristics. The accuracy is evaluated by comparison with the DRR estimated directly from the room impulse response. The difference is typically 5 dB. For two rooms, the variation of the DRR estimate with source-to-receiver position is also shown. The method is blind in the sense that it is virtually independent of the signal generated by a single sound source.

Psychoacoustic evaluation of multichannel reproduced sounds using binaural synthesis and spherical beamforming

The binaural auralization of a 3D sound field using spherical-harmonics beamforming (SHB) techniques was investigated and compared with the traditional method using a head-and-torso simulator (HATS). The new procedure was verified by comparing simulated room impulse responses with measured ones binaurally. The objective comparisons show that there is good agreement in the frequency range between 0.1 and 6.4 kHz. A listening experiment was performed to validate the SHB method subjectively and to compare it to the HATS method. Two musical excerpts, pop and classical, were used. Subjective responses were collected in two head rotation conditions (fixed and rotating) and six spatial reproduction modes, including phantom mono, stereo, and surround sound. The results show that subjective scales of width, spaciousness, and preference based on the SHB method were similar to those obtained for the HATS method, although the width and spaciousness of the stimuli processed by the SHB method were judged slightly higher than the ones using the HATS method in general. Thus, binaural synthesis using SHB may be a useful tool to reproduce a 3D sound field binaurally, while saving considerably on measurement time because head rotation can be simulated based on a single recording.

Subjective Response to Foot-Fall Noise, Including Localization of the Source Position

Although an impact noise level is objectively evaluated the same according to current standards, a lightweight floor structure is often subjectively judged more annoying than a heavy homogeneous structure. The hypothesis of the present investigation is that the subjective judgment of impact noise is more annoying if the source position can be localized; lightweight structures have a more localized radiation than heavy structures. For the heavy structures the reverberant vibration field is dominant, therefore having a distributed radiation. A listening test is used to assess the subjective annoyance, using simulated binaural room impulse responses, with sources being a moving point source or a non-moving surface source, and rooms being a room with a reverberation time of 0.5 s or an anechoic room. The paper concludes that no strong effect of the source localization on the annoyance can be found.

Perceptual validation of virtual room acoustics: Sound localisation and speech understanding

The reliability of algorithms for room acoustic simulations has often been confirmed on the basis of the verification of predicted room acoustical parameters. This paper presents a complementary perceptual validation procedure consisting of two experiments, respectively dealing with speech intelligibility, and with sound source front–back localisation. The evaluated simulation algorithm, implemented in software ODEON®, is a hybrid method that is based on an image source algorithm for the prediction of early sound reflection and on ray-tracing for the later part, using a stochastic scattering process with secondary sources. The binaural room impulse response (BRIR) is calculated from a simulated room impulse response where information about the arriving time, intensity and spatial direction of each sound reflection is collected and convolved with a measured Head Related Transfer Function (HRTF). The listening stimuli for the speech intelligibility and localisation tests are auralised convolutions of anechoic sound samples with measured and simulated BRIRs. Perception tests were performed with human subjects in two acoustical environments, i.e. an anechoic and reverberant room, by presenting the stimuli to subjects in a natural way, and via headphones by using two non-individualized HRTFs (artificial head and hearing aids placed on the ears of the artificial head) of both a simulated and a real room. Very good correspondence is found between the results obtained with simulated and measured BRIRs, both for speech intelligibility in the presence of noise and for sound source localisation tests. In the anechoic room an increase in speech intelligibility is observed when noise and signal are presented from sources located at different angles. This improvement is not so evident in the reverberant room, with the sound sources at 1-m distance from the listener. Interestingly, the performance of people for front–back localisation is better in the reverberant room than in the anechoic room. The correlation between people’s ability for sound source localisation on one hand, and their ability for recognition of binaurally received speech in reverberation on the other hand, is found to be weak.

Chinese speech intelligibility at different speech sound pressure levels and signal-to-noise ratios in simulated classrooms

The speech intelligibility in classroom can be influenced by background-noise levels, speech sound pressure level (SSPL), reverberation time and signal-to-noise ratio (SNR). The relationship between SSPL and subjective Chinese Mandarin speech intelligibility and the effect of different SNRs on Chinese Mandarin speech intelligibility in the simulated classroom were investigated through room acoustical simulation, auralisation technique and subjective evaluation. Chinese speech intelligibility test signals recorded in anechoic chamber were convolved with the simulated binaural room impulse responses, and then reproduced through the headphone by different SSPLs and SNRs. The results show that Chinese Mandarin speech intelligibility scores increase with increasing of SSPLs and SNRs within a certain range in simulated classrooms. Chinese Mandarin speech intelligibility scores have no significant difference with SNRs of no less than 15 dBA under the same reverberation time condition.

Relationship between Chinese speech intelligibility and speech transmission index in rooms using dichotic listening

Speech intelligibility (SI) is an important index for the design and assessment of speech purpose hall. The relationship between Chinese speech intelligibility scores in rooms and speech transmission index (STI) under diotic listening condition was studied using monaural room impulse responses obtained from the room acoustical simulation software Odeon in previous paper. The present study employs the simulated binaural room impulse responses and auralization technique to obtain the subjective Chinese speech intelligibility scores using rhyme test. The relationship between Chinese speech intelligibility scores and STI is built and validated in rooms using dichotic (binaural) listening. The result shows that there is a high correlation between Chinese speech intelligibility scores and STI using dichotic listening. The relationship between Chinese speech intelligibility scores and STI under diotic and dichotic listening conditions is also analyzed. Compared with diotic listening, dichotic (binaural) listening (an actual listening situation) can improve 2.7 dB signal-to-noise ratio for Mandarin Chinese speech intelligibility. STI method can predict and evaluate the speech intelligibility for Mandarin Chinese in rooms for dichotic (binaural) listening.

Binaural auralization based on spherical-harmonics beamforming

The binaural auralization of a 3D sound field using spherical-harmonics beamforming (SHB) techniques was investigated and compared with the traditional method using a dummy head. The new procedure was verified by comparing simulated room impulse responses with directly measured ones both monaurally and binaurally. The objective comparisons show that there is good agreement in the frequency range between 0.1 to 6.4 kHz. Psychoacoustic attributes of multi-channel reproduced sounds were measured in a listening experiment to validate the method subjectively. The results show that subjective ratings of the width, spaciousness and preference of different audio reproduction modes auralized based on SHB were not significantly different from those obtained for dummy head measurements. Thus binaural synthesis using SHB may be a useful tool to reproduce a 3D sound field binaurally while saving considerably on measurement time because head rotation can be simulated based on a single recording.

Relationship between Chinese speech intelligibility and speech transmission index using diotic listening

The speech intelligibility in rooms is evaluated using the room impulse responses obtained from the room acoustical simulation software ODEON. The simulated room impulse responses are first convolved with the speech intelligibility test signals recorded in an anechoic chamber, then reproduced through the earphone. The subjective Chinese speech intelligibility scores are obtained and the relationship between Chinese speech intelligibility scores and speech transmission index (STI) is built and validated. The result shows that there is high correlation between Chinese speech intelligibility scores and STI. The STI method can predict and evaluate the speech intelligibility for Mandarin Chinese without changes in the algorithm of the weighting values for diotic listening in rooms.

Feasibility of subjective speech intelligibility assessment based on auralization

Subjective speech intelligibility can be assessed by speech recorded in an anechoic chamber and then convolved with room impulse responses that can be created by acoustic simulation. The speech intelligibility (SI) assessment based on auralization was validated in three rooms. The articulation scores obtained from simulated sound field were compared with the ones from measured sound field and from direct listening in rooms. Results show that the speech intelligibility prediction based on auralization technique with simulated binaural room impulse responses (BRIRs) is in agreement with reality and results from measured BRIRs. When this technique is used with simulated and measured monaural room impulse responses (MRIRs), the predicted results underestimate the reality. It has been shown that auralization technique with simulated BRIRs is capable of assessing subjective speech intelligibility of listening positions in the room.

A microphone array processing technique for speech enhancement in a reverberant space

In this paper, a new microphone array processing technique is proposed for blind dereverberation of speech signals affected by room acoustics. It is based on the separate processing of the minimum-phase and all-pass components of delay-steered multi-microphone signals. The minimum-phase components are processed in the cepstrum-domain, where spatial averaging followed by low-time filtering is applied. The all-pass components, which contain the source location information, are processed in the frequency-domain by performing spatial averaging and by retaining only the all-pass component of the resulting output. The underlying motivation for the new processor is to use spatio-temporal processing over a single set of synchronous speech segments from several microphones to reconstruct the source speech, such that it is applicable to practical time-variant acoustic environments. Simulated room impulse responses are used to evaluate the new processor and to compare it to a conventional beamformer. Significant improvements in array gain and important reductions of reverberation in listening tests are observed.