Segmentation into audio document speakers: A new approach based on the one- class support vector methods
Bibliographic record
Abstract
With recent and continued increases in the number of available sound archives (radio, TV, Web,...), effective methods must be established to facilitate the process of searching for information within massive databases. Of less complexity than the original sound file but nevertheless containing a summary of important information pertaining to the signal, text files (index files) are linked to the digital sound files. An example of relevant information found in the text file is as follows: 45 minutes of speech, 1 minute of music, 10 speakers (6 men and 4 women). These index files, stored with the original signal, will contribute considerably to the information retrieval process, allowing an immediate and direct access to the information sought. If one would like to know who speaks and when in a sound file, the index key is hence the speaker. A preliminary stage of a speaker indexing system is speaker diarization. State-of-the-art speaker diarization techniques require two main steps: speaker turn detection which consists of detecting speaker turn times, that is boundaries of audio file segments where only one speaker is present, followed by a clustering step which consists of labelling the previous segments in terms of speakers. These two stages require a metric to be defined in order to compare and groups speech segments. This paper presents a novel approach for the speaker diarization of audio recordings. The proposed approach uses a metric based on one-class Support Vector Machines (SVM-I), introduced recently by one of the authors, for the speaker change detection and clustering tasks. Through many experiments using two databases of broadcast recordings, we demonstrate the relevance and superiority of this approach compared to the traditional method based on the generalized likelihood ratio using bayesian information criterion (RVG-BIC).
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 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".