AI Paper Atlas

Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning

Towards Zero-Shot Learning for Automatic Phonemic Transcription

Xinjian Li, Siddharth Dalmia, David Mortensen, Juncheng Li, Alan Black, Florian Metze

AAAI 2020 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Robust fine-tuning of zero-shot models · 27 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use articulatory-attribute prediction and a language-specific phoneme signature matrix instead of a shared phoneme output inventoryMultilingual acoustic model · 7-language averagePER (%): 66.9 -> 59.2Improved
Use articulatory-attribute prediction and a language-specific phoneme signature matrixMultilingual acoustic model · 7-language averageSubstitution error (%): 40.2 -> 31.9Improved
Use articulatory-attribute prediction to infer unseen phonemesMultilingual acoustic model · 7-language average, unseen phonemesPER (%): 100.0 -> 89.8Improved
Use articulatory-attribute predictionMultilingual acoustic model · 7-language average, seen phonemesPER (%): 64.2 -> 57.0Improved
Increase the number of training languages (ablation)UPM and multilingual baseline · 7-language averagePER (%)Improved
Use more training languages with UPM rather than the multilingual baseline (ablation)UPM versus multilingual baseline · 7-language averagePER gap: 5.0 to 9.3Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy