Commit
•
40b9d12
1
Parent(s):
03b27b4
Update README.md
Browse files
README.md
CHANGED
|
@@ -15,10 +15,10 @@ pipeline_tag: audio-classification
|
|
| 15 |
|
| 16 |
We build a CTC-based phoneme recognition model using wav2vec 2.0 (W2V2) for children under 4-year-old. We use three-level fine-tuning to gradually reduce age mismatch between adult phonetics to child phonetics.
|
| 17 |
|
| 18 |
-
**W2V2-Libri100h**: We first fine-tune W2V2-Base pretrained on unlabeled 960h adult speech corpus with IPA phone sequences.
|
| 19 |
-
**W2V2-MyST**: We then fine-tune W2V2-Libri100h using [My Science Tutor](https://boulderlearning.com/products/myst/) corpus (consists of conversational speech of students between the third and fifth grades with a virtual tutor).
|
| 20 |
-
**W2V2-Libri100h-Pro (two-level fine-tuning)**: We fine-tune W2V2-Libri100h using [Providence](https://phonbank.talkbank.org/access/Eng-NA/Providence.html) corpus (consists of longititude audio of 6 English-speaking children aged from 1-4 years interacting with their mothers at home) on phoneme sequences.
|
| 21 |
-
**W2V2-MyST-Pro (three-level fine-tuning)**: Similar as W2V2-Libri100h-Pro, we fine-tune W2V2-MyST using Providence on phoneme sequences.
|
| 22 |
|
| 23 |
We show W2V2-MyST-Pro is helpful for improving children's vocalization classification task on two corpus, including [Rapid-ABC](https://openaccess.thecvf.com/content_cvpr_2013/html/Rehg_Decoding_Childrens_Social_2013_CVPR_paper.html) and [BabbleCor](https://osf.io/rz4tx/).
|
| 24 |
|
|
|
|
| 15 |
|
| 16 |
We build a CTC-based phoneme recognition model using wav2vec 2.0 (W2V2) for children under 4-year-old. We use three-level fine-tuning to gradually reduce age mismatch between adult phonetics to child phonetics.
|
| 17 |
|
| 18 |
+
- **W2V2-Libri100h**: We first fine-tune W2V2-Base pretrained on unlabeled 960h adult speech corpus with IPA phone sequences.
|
| 19 |
+
- **W2V2-MyST**: We then fine-tune W2V2-Libri100h using [My Science Tutor](https://boulderlearning.com/products/myst/) corpus (consists of conversational speech of students between the third and fifth grades with a virtual tutor).
|
| 20 |
+
- **W2V2-Libri100h-Pro (two-level fine-tuning)**: We fine-tune W2V2-Libri100h using [Providence](https://phonbank.talkbank.org/access/Eng-NA/Providence.html) corpus (consists of longititude audio of 6 English-speaking children aged from 1-4 years interacting with their mothers at home) on phoneme sequences.
|
| 21 |
+
- **W2V2-MyST-Pro (three-level fine-tuning)**: Similar as W2V2-Libri100h-Pro, we fine-tune W2V2-MyST using Providence on phoneme sequences.
|
| 22 |
|
| 23 |
We show W2V2-MyST-Pro is helpful for improving children's vocalization classification task on two corpus, including [Rapid-ABC](https://openaccess.thecvf.com/content_cvpr_2013/html/Rehg_Decoding_Childrens_Social_2013_CVPR_paper.html) and [BabbleCor](https://osf.io/rz4tx/).
|
| 24 |
|