Bibliographic record
Abstract
This work introduces new binaural beamforming algorithms for hearing aids, with a robustness to errors in the estimate of the target speaker direction of arrival (DOA) and a good trade-off between noise reduction and preservation of the noise/interferers spatial impression. Three robust designs are proposed, and their robustness is confirmed by simulation results. These robust designs are a combination of binaural and monaural beamformers using two different microphone configurations: one for low frequency components and one for high frequency components. The robust designs are also found to be robust to mismatch between the anechoic propagation models used for the beamformers designs and the reverberant propagation models used to generate the signals at the microphones in the simulations. To preserve the binaural cues of the noise/interferers in the binaural beamformer outputs, a method based on a mixing/selection of different available binaural signals is proposed, using a classification from the phase and magnitude of a complex coherence function. This method is added as a post processor to the beamforming designs robust to target DOA mismatch. Simulation results show that the resulting mixed binaural output signals have a good binaural cues preservation level that outperform the benchmark design, with significant noise reduction and low target distortion. Since knowledge of source DOAs is important for beamforming noise reduction, a beamformer-based broadband multi-source DOA detection system is also developed in the thesis, using information from different frequencies or sub‐bands to obtain global estimates of sources DOAs. Simulation results shows that using one beamformer on each side is capable of detecting the DOAs of active sources under several acoustic scenarios, including scenarios with one, two, or three sources, and with or without the presence of some level of diffuse noise.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".