Better model and decoding methods for automatic speech recognition
Bibliographic record
Abstract
Different ways to improve the accuracy and performance of automatic speech recognition (ASR) systems are examined. Earlier research in the INRS group concentrated on improvements in adaptation techniques, to try to reduce the mismatch between system training and operating conditions. More recently, a concentration was made on improvements to the modeling and decoding methodology of the ASR systems. A two-level syllable model-based decoding approach is proposed here. At the first level, a syllable model-based decoding is performed to segment the utterances into syllable segments, and to identify which syllable group this segment belongs to. Then, at the second level, each segment is rescored using the phoneme model, where the possible phoneme sequences are constrained by the syllables in that syllable group. A new heuristic score is proposed to be added in hidden Markov model (HMM) decoding that indicates the degree of competition among different HMM states. Speech features are obtained from the posterior of this HMM approach and these features are used to further improve recognition results. The state evolution tracks are incorporated as features in an HMM-based recognizer. Experiments were carried out on a phoneme classification task (TIMIT) and these methods were found to show improvements compared to a baseline performance. [Work supported by NSERC-Canada and Prompt-Quebec.]
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 | 0.004 | 0.009 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.009 | 0.010 |
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".