Spaces:
Runtime error
Runtime error
Updating description
Browse files
app.py
CHANGED
@@ -332,12 +332,12 @@ interface = gr.Interface(
|
|
332 |
To efficiently compute text difficulty, a Distil-Bert pre-trained model is fine-tuned for regression using The CommonLit Ease of Readability (CLEAR)
|
333 |
Corpus. https://educationaldatamining.org/EDM2021/virtual/static/pdf/EDM21_paper_35.pdf This dataset contains over 110,000 pairwise comparisons of
|
334 |
~1100 teachers responded to the question, "Which text is easier for students to understand?". This model is trained end-end (regression layer down to
|
335 |
-
the first attention layer to ensure the best performance- Merchant et al. 2020
|
336 |
|
337 |
Speech Pronunciaion Scoring: The Wave2Vec 2.0 model is utilized to convert audio into text in real-time. The model predicts words or phonemes (smallest
|
338 |
-
unit of speech distinguishing one word (or word element) from another) from the input audio
|
339 |
-
pronunciation
|
340 |
-
pass this audio through Wave2Vec to get the inferred intended words. We measure the loss as the Levenshtein distance between the target and actual transcripts-
|
341 |
the Levenshtein distance between two words is the minimum number of single-character edits required to change one word into the other.
|
342 |
|
343 |
Lexical Diversity Score: The lexical diversity score is computed by taking the ratio of unique similar words to total similar words squared. The similarity is computed
|
|
|
332 |
To efficiently compute text difficulty, a Distil-Bert pre-trained model is fine-tuned for regression using The CommonLit Ease of Readability (CLEAR)
|
333 |
Corpus. https://educationaldatamining.org/EDM2021/virtual/static/pdf/EDM21_paper_35.pdf This dataset contains over 110,000 pairwise comparisons of
|
334 |
~1100 teachers responded to the question, "Which text is easier for students to understand?". This model is trained end-end (regression layer down to
|
335 |
+
the first attention layer) to ensure the best performance- Merchant et al. 2020
|
336 |
|
337 |
Speech Pronunciaion Scoring: The Wave2Vec 2.0 model is utilized to convert audio into text in real-time. The model predicts words or phonemes (smallest
|
338 |
+
unit of speech distinguishing one word (or word element) from another) from the user input audio. Due to the nature of the model, users with poor
|
339 |
+
pronunciation receive inaccurate translations. This project attempts to score pronunciation by asking a user to read a target excerpt into the microphone. We then
|
340 |
+
pass this audio through Wave2Vec 2.0 to get the inferred intended words. We measure the loss as the Levenshtein distance between the target and actual transcripts-
|
341 |
the Levenshtein distance between two words is the minimum number of single-character edits required to change one word into the other.
|
342 |
|
343 |
Lexical Diversity Score: The lexical diversity score is computed by taking the ratio of unique similar words to total similar words squared. The similarity is computed
|