Research on Automatic Speech Recognition for dysarthric speech
The effect of speech pathology on automatic speaker verification
Animate a 3D tongue based on a physics simulation of muscles and pressure.
Hyb-DysNet: hybrid feature fusion (eGeMAPSv02 + Wav2Vec2-XLS-R + Librosa) and soft-voting ensemble for dysarthria severity classification in ALS patients.
Cross-platform application, developed in Flutter, for mobile phones, watches and tablets, for an Augmentative and Alternative Communication (AAC) application, intended to be used by people suffering from speech impairments such as apraxia and/or dysarthria of speech.
User guide, tutorial videos, presentations, logos and artwork, for an Augmentative and Alternative Communication (AAC) application, intended to be used by people suffering from speech impairments such as apraxia and/or dysarthria of speech.
Web page for the project, privacy policy and terms and conditions, for an Augmentative and Alternative Communication (AAC) application, intended to be used by people suffering from speech impairments such as apraxia and/or dysarthria of speech.
LoRA fine-tuning of OpenAI Whisper for dysarthric speech recognition using TORGO.
Local, privacy-first speech biomarker platform for longitudinal neurological monitoring (Parkinson's, dysarthria) — descriptive acoustic markers, no diagnosis
A project to classify dysarthric and non-dysarthric speech using deep learning techniques.
Offline, personalized speech assistant for dysarthria users. Fine-tunes Gemma 4 on a single user's voice and runs on-device via LiteRT-LM.
A multi-modal acoustic-aware fusion architecture for dysarthria detection and severity classification. Combines LoRA-adapted wav2vec 2.0 with classical MFCC features in a dual-pathway network, grounding foundation models in deterministic acoustic physics for clinical interpretability.
Audio-visual speech recognition for the UASpeech dysarthric-speech dataset. Whisper + AV-HuBERT fusion in PyTorch Lightning, with audio-only and AV training, dataset prep, and per-intelligibility metrics.
Korean jamo-level STT for dysarthria assessment support with DeepSpeech2, Simple-Attention, and Transformer model comparison.
A deep learning project for automated speech and language assessment in post-stroke aphasia and dysarthria.
A voice assistant for people whose speech other assistants mishear: three big guesses, and nothing happens until you choose. Built for Alexa+ on AWS, with Amazon Nova, an MCP confirm gate and a measured evaluation on dysarthric speech.
Multilingual Audio-Based detection and voice restoration of Dysarthria using machine learning and signal processing techniques.
This is the code belonged to the team working on the project in CCCN lab, Bangkok. This code is open to the public use with permission to primary author. Citation to the attached paper is required for all further publications.
Detecting dysarthria from speech audio using deep learning - Masters Dissertation
Edge-native real-time speech restoration & assistive communication engine for motor-speech impairments.
To associate your repository with the dysarthria topic, visit your repo's landing page and select "manage topics."