Research map / Probabilistic and scientific modeling / Variational and generative sampling
EA-VAE: Learning to Reconstruct Dysarthric Speech via Variational Autoencoder with Encoding Alignment
Daipeng Zhang, Wenhuan Lu, Xianghu Yue, Hongcheng Zhang, Jianguo Wei
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add Distribution Alignment Module (KL between dysarthric and normal priors) (ablation) | EA-VAE · UASpeech B2 average | WER (%): 94.46 -> 62.19 | Improved |
| add Distribution Alignment Module (ablation) | EA-VAE · UASpeech B2 average | MOS: 1.66 -> 4.48 | Improved |
| add Embedding Alignment Module (ablation) | EA-VAE · UASpeech B2 average | WER (%): 71.75 -> 62.19 | Improved |
| add Embedding Alignment Module (ablation) | EA-VAE · UASpeech B2 average | MOS: 3.56 -> 4.48 | Improved |
| add Duration Alignment Predictor (ablation) | EA-VAE · UASpeech B2 average | WER (%): 63.43 -> 62.19 | Improved |
| add Duration Alignment Predictor (ablation) | EA-VAE · UASpeech B2 average | MOS: 4.13 -> 4.48 | Improved |
| time-stretch dysarthric audio so Td >= Tn (vs Td < Tn) (ablation) | EA-VAE · UASpeech B2 average | WER (%): 83.60 -> 62.19 | Improved |
| time-stretch dysarthric audio so Td >= Tn (vs Td < Tn) (ablation) | EA-VAE · UASpeech B2 average | MOS: 1.23 -> 4.48 | Improved |
| use EA-VAE instead of Unit-DSR | dysarthric speech reconstruction · UASpeech B2 average | WER (%): 65.08 -> 62.19 | Improved |
| use EA-VAE instead of ASR-TTS on speaker F02 MOS | dysarthric speech reconstruction · UASpeech F02 (Low) | MOS: 4.02 vs Unit-DSR 4.42 | Mixed |
Closest papers by meaning
- Continuous Autoregressive Modeling with Stochastic Monotonic Alignment for Speech Synthesis (ICLR 2025 · similarity 0.84)
- SSDM: Scalable Speech Dysfluency Modeling (NeurIPS 2024 · similarity 0.81)
- SSC-VAE: Structured Sparse Coding Based Variational Autoencoder for Detail Preserved Image Reconstruction (AAAI 2025 · similarity 0.81)
- Separate and Reconstruct: Asymmetric Encoder-Decoder for Speech Separation (NeurIPS 2024 · similarity 0.81)
- TaDiCodec: Text-aware Diffusion Speech Tokenizer for Speech Language Modeling (NeurIPS 2025 · similarity 0.81)
- SECodec: Structural Entropy-based Compressive Speech Representation Codec for Speech Language Models (AAAI 2025 · similarity 0.80)
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.
