Automated Mandarin tone classification using deep neural networks trained on a large speech dataset
Bibliographic record
Abstract
A tonal language is one in which the speaker’s intonation modifies the meaning of a word. In this work, we perform a rigorous analysis of intonation changes, or pitch contours, produced by native Mandarin speakers to predict the tone-contour type. Pitch contours are estimated using a number of different methods, also measuring each contour’s Mel-Frequency Cepstral Coefficients (MFCCs). The dataset used was autonomously generated from the Aishell open-source Mandarin speech corpus. Each sample was aligned with its transcript using Montreal Forced Alignment and segmented into individual words. The resulting corpus covers 11 topic domains, spoken by 400 individuals. Separate development, training, and testing datasets are created to ensure the integrity of our results. Pitch contours and their MFCCs are exposed to a number of machine learning techniques including clustered, regression, and traditional Deep Neural Network (DNN) approaches. MFCCs are additionally processed using convolutional neural networks. The models are used to predict the corresponding tone for a contour. Our work seeks to determine which intonation representations perform optimally for machine learning tasks. The tool is used to provide audio and visual feedback to learners of tonal languages. [Work supported by RPI Seed Grant and CISL].
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".