Insights into optimal phonemic compression from a computational model of the auditory periphery
Bibliographic record
Abstract
Phonemic compression schemes for hearing aids have thus far been developed and evaluated based on perceptual criteria such as speech intelligibility, sound comfort, and loudness equalization. Finding compression parameters that optimize all of these perceptual metrics has proved difficult. The goal of this study was to nd optimal single-band gain adjustments based on the response of auditory-nerve bers to speech. Sentences from the TIMIT database were processed by either the NAL-R or the DSL ampli cation scheme, and deviations from these linear prescriptions were obtained by adjusting the overall gain from 40 dB below to 40 dB above the prescribed gains in 5 dB steps. Neural responses were obtained using the cat auditory-periphery model of Zilany and Bruce (2006, 2007). Sentences were analyzed on a phone by phone basis to nd the gain adjustment that minimized the difference in neural response to the amplified phone in the impaired model and the unampli ed phone in the normal model. The optimal gain adjustments were found to depend on whether the error metric included the spike timing information of the neural responses (i.e., a time resolution of several microseconds) or just the mean ring rates (i.e., a time resolution of several milliseconds). To optimize the mean ring rates, gain adjustments on the order of +10 dB were required above the prescribed linear gains in general. In contrast, gain adjustments on the order of 10 dB or more below the prescribed linear gains tended to optimize the responses including spike timing information. Wide dynamic range compression appears to be more bene cial in optimizing the spike timing information than the mean rate information. These results motivate the development of novel nonlinear amplification schemes that simultaneously optimize both spike-timing and mean-rate neural representations.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
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".