Reduction of Subjective Listening Effort for TV Broadcast Signals With Recurrent Neural Networks

Westhausen, Nils L.; Hüber, Rainer; Baumgartner, H. Leroy; Sinha, Ragini; Rennies, Jan; Meyer, Bernd T.

doi:10.1109/taslp.2021.3126931

Cited by 11 publications

(4 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The data obtained here hardly ever reach this value, except for a few individual tracks, with mean values in the low 20 th ies. This suggests that we are looking to a less than a "slight improvement", on average, which is in line with algorithmic approaches that provides SNR enhancement [23]. Anyway, the aim of the study was to improve speech quality from flat TVsets in speech excerpts with very low noise.…”

Section: Resultsmentioning

confidence: 72%

“…Many issues still are not fully settled for audio processing based on SNR enhancement in the broadcasting field. In particular, the source separation and subsequent remixing at a higher SNR relative to the original mix [23] is one of the most promising and recently proposed method for improving the perceived speech quality and reducing the listening effort in TV broadcasting. This method reaches the goal of providing easier speech perception while preserving the original sound atmosphere as much as possible.…”

Section: Resultsmentioning

confidence: 99%

“…Recently, the speech enhancement for hearing-impaired listeners has been improved thanks to a system based on sound source separation and recurrent neural networks (RNN). The separation of speech from background signals and the remixing at a higher signal-to-noise ratio (SNR) brought to a reduction of the listening effort and an increase in the perceived sound quality on real TV-broadcast material [23]. Such approaches are potentially more powerful than a simple equalization because they can actually provide a SNR enhancement.…”

Section: B State Of the Art On Tv Listening Experience Of Hearing-imp...mentioning

confidence: 99%

See 2 more Smart Citations

Speech Quality Improvement of TV-Sets for Hearing-Impaired Older Adults

Astolfi

Riente

Albera

et al. 2023

IEEE Trans. on Broadcast.

View full text Add to dashboard Cite

Flat TV-sets sacrifice space for built-in loudspeakers and this implies a degradation of the speech quality, especially for hearing-impaired older adults. Many elderly people report having difficulty understanding speech on TV, even with hearing aids. In the study, a Transfer Function (TF) that improves the listening experience for the above categories of listeners is validated for Rai, the Italian TV broadcasting company. A Digital Audio Optimizer dynamically equalizes the sound level, in real-time, before the transmission to the broadcasting tower. It implements a TF that amplifies the frequency range between 1 kHz to 4 kHz, which is more important for speech intelligibility, with a particular boosting of the one-third octave band of 4 kHz. Subjective tests of the proposed TF have been carried out in a laboratory in compliance with the standard ITU-R BS.1116-3 with 31 hearing-impaired older adults and a commercial flat-screen TV-set. Results showed a statistically significant slight improvement in the perceived audio quality of 24.3%, over three genres, i.e., "Speech", "Singing and Music" and "Sport". This work follows a previous study on 72 normal-hearing subjects where the same methodology was applied, which involved three commercial flat-screen TV-sets and the same genres. The perceived audio quality was judged improved by the 25.3% on average.

show abstract

Section: Resultsmentioning

confidence: 72%

Section: Resultsmentioning

confidence: 99%

Section: B State Of the Art On Tv Listening Experience Of Hearing-imp...mentioning

confidence: 99%

See 1 more Smart Citation

Speech Quality Improvement of TV-Sets for Hearing-Impaired Older Adults

Astolfi

Riente

Albera

et al. 2023

IEEE Trans. on Broadcast.

View full text Add to dashboard Cite

show abstract

“…This could be achieved through suppression of only the late part of the reverberation as performed in [44] compared to the full dereverberation performed in this study. Alternatively, the output signal could be remixed with the unprocessed signal in a certain ratio [45], which increases awareness for the acoustic scene and can mask artifacts of the source separation procedure.…”

Section: Effect Of Model Size and Groupingmentioning

confidence: 99%

Binaural Multichannel Blind Speaker Separation With a Causal Low-Latency and Low-Complexity Approach

Westhausen,

Meyer

2024

IEEE Open J. Signal Process.

Self Cite

View full text Add to dashboard Cite

In this article, we introduce a causal low-latency low-complexity approach for binaural multichannel blind speaker separation in noisy reverberant conditions. The model, referred to as Group Communication Binaural Filter and Sum Network (GCBFSnet) predicts complex filters for filter-and-sum beamforming in the time-frequency domain. We apply Group Communication (GC), i.e., latent model variables are split into groups and processed with a shared sequence model with the aim of reducing the complexity of a simple model only containing one convolutional and one recurrent module. With GC we are able to reduce the size of the model by up to 83% and the complexity up to 73% compared to the model without GC, while mostly retaining performance. Even for the smallest model configuration, GCBFSnet matches the performance of a low-complexity TasNet baseline in most metrics despite the larger size and higher number of required operations of the baseline.INDEX TERMS Binaural, low-latency, multi-channel, real-time, speaker-separation.

show abstract