@article{52958,
  author       = {{Boeddeker, Christoph and Subramanian, Aswin Shanmugam and Wichern, Gordon and Haeb-Umbach, Reinhold and Le Roux, Jonathan}},
  issn         = {{2329-9290}},
  journal      = {{IEEE/ACM Transactions on Audio, Speech, and Language Processing}},
  keywords     = {{Electrical and Electronic Engineering, Acoustics and Ultrasonics, Computer Science (miscellaneous), Computational Mathematics}},
  pages        = {{1185--1197}},
  publisher    = {{Institute of Electrical and Electronics Engineers (IEEE)}},
  title        = {{{TS-SEP: Joint Diarization and Separation Conditioned on Estimated Speaker Embeddings}}},
  doi          = {{10.1109/taslp.2024.3350887}},
  volume       = {{32}},
  year         = {{2024}},
}

@article{35602,
  abstract     = {{Continuous Speech Separation (CSS) has been proposed to address speech overlaps during the analysis of realistic meeting-like conversations by eliminating any overlaps before further processing.
CSS separates a recording of arbitrarily many speakers into a small number of overlap-free output channels, where each output channel may contain speech of multiple speakers.
This is often done by applying a conventional separation model trained with Utterance-level Permutation Invariant Training (uPIT), which exclusively maps a speaker to an output channel, in sliding window approach called stitching.
Recently, we introduced an alternative training scheme called Graph-PIT that teaches the separation network to directly produce output streams in the required format without stitching.
It can handle an arbitrary number of speakers as long as never more of them overlap at the same time than the separator has output channels.
In this contribution, we further investigate the Graph-PIT training scheme.
We show in extended experiments that models trained with Graph-PIT also work in challenging reverberant conditions.
Models trained in this way are able to perform segment-less CSS, i.e., without stitching, and achieve comparable and often better separation quality than the conventional CSS with uPIT and stitching.
We simplify the training schedule for Graph-PIT with the recently proposed Source Aggregated Signal-to-Distortion Ratio (SA-SDR) loss.
It eliminates unfavorable properties of the previously used A-SDR loss and thus enables training with Graph-PIT from scratch.
Graph-PIT training relaxes the constraints w.r.t. the allowed numbers of speakers and speaking patterns which allows using a larger variety of training data.
Furthermore, we introduce novel signal-level evaluation metrics for meeting scenarios, namely the source-aggregated scale- and convolution-invariant Signal-to-Distortion Ratio (SA-SI-SDR and SA-CI-SDR), which are generalizations of the commonly used SDR-based metrics for the CSS case.}},
  author       = {{von Neumann, Thilo and Kinoshita, Keisuke and Boeddeker, Christoph and Delcroix, Marc and Haeb-Umbach, Reinhold}},
  issn         = {{2329-9290}},
  journal      = {{IEEE/ACM Transactions on Audio, Speech, and Language Processing}},
  keywords     = {{Continuous Speech Separation, Source Separation, Graph-PIT, Dynamic Programming, Permutation Invariant Training}},
  pages        = {{576--589}},
  publisher    = {{Institute of Electrical and Electronics Engineers (IEEE)}},
  title        = {{{Segment-Less Continuous Speech Separation of Meetings: Training and Evaluation Criteria}}},
  doi          = {{10.1109/taslp.2022.3228629}},
  volume       = {{31}},
  year         = {{2023}},
}

@article{11861,
  abstract     = {{In this contribution we present a theoretical and experimental investigation into the effects of reverberation and noise on features in the logarithmic mel power spectral domain, an intermediate stage in the computation of the mel frequency cepstral coefficients, prevalent in automatic speech recognition (ASR). Gaining insight into the complex interaction between clean speech, noise, and noisy reverberant speech features is essential for any ASR system to be robust against noise and reverberation present in distant microphone input signals. The findings are gathered in a probabilistic formulation of an observation model which may be used in model-based feature compensation schemes. The proposed observation model extends previous models in three major directions: First, the contribution of additive background noise to the observation error is explicitly taken into account. Second, an energy compensation constant is introduced which ensures an unbiased estimate of the reverberant speech features, and, third, a recursive variant of the observation model is developed resulting in reduced computational complexity when used in model-based feature compensation. The experimental section is used to evaluate the accuracy of the model and to describe how its parameters can be determined from test data.}},
  author       = {{Leutnant, Volker and Krueger, Alexander and Haeb-Umbach, Reinhold}},
  issn         = {{2329-9290}},
  journal      = {{IEEE/ACM Transactions on Audio, Speech, and Language Processing}},
  keywords     = {{computational complexity, reverberation, speech recognition, automatic speech recognition, background noise, clean speech, computational complexity, energy compensation, logarithmic mel power spectral domain, mel frequency cepstral coefficients, microphone input signals, model-based feature compensation schemes, noisy reverberant speech automatic recognition, noisy reverberant speech features, reverberation, Atmospheric modeling, Computational modeling, Noise, Noise measurement, Reverberation, Speech, Vectors, Model-based feature compensation, observation model for reverberant and noisy speech, recursive observation model, robust automatic speech recognition}},
  number       = {{1}},
  pages        = {{95--109}},
  title        = {{{A New Observation Model in the Logarithmic Mel Power Spectral Domain for the Automatic Recognition of Noisy Reverberant Speech}}},
  doi          = {{10.1109/TASLP.2013.2285480}},
  volume       = {{22}},
  year         = {{2014}},
}