Spaces:
Running
Running
arunasrivastava
commited on
Commit
·
9926870
1
Parent(s):
c8d97f4
update panphon
Browse files- app.py +7 -0
- requirements.txt +1 -1
app.py
CHANGED
@@ -153,6 +153,13 @@ with gr.Blocks(css="""
|
|
153 |
- **PER (Phoneme Error Rate)**: The Levenshtein distance calculated between phoneme sequences of the predicted and actual transcriptions.
|
154 |
- **PWED (Phoneme Weighted Edit Distance)**: A measure of the edit distance between the predicted and actual phoneme sequences, weighted by the phonemic feature distance. Feature vectors provided by panphon library
|
155 |
""")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
156 |
with gr.Tabs():
|
157 |
with gr.TabItem("🏆 Leaderboard"):
|
158 |
leaderboard_html = gr.HTML(create_html_table(format_leaderboard_df(load_leaderboard_data())))
|
|
|
153 |
- **PER (Phoneme Error Rate)**: The Levenshtein distance calculated between phoneme sequences of the predicted and actual transcriptions.
|
154 |
- **PWED (Phoneme Weighted Edit Distance)**: A measure of the edit distance between the predicted and actual phoneme sequences, weighted by the phonemic feature distance. Feature vectors provided by panphon library
|
155 |
""")
|
156 |
+
gr.Markdown("""
|
157 |
+
## Test Set Information
|
158 |
+
The test set used for evaluation is from the [TIMIT speech corpus](https://www.kaggle.com/datasets/mfekadu/darpa-timit-acousticphonetic-continuous-speech). The TIMIT corpus is a widely used dataset for speech recognition research.
|
159 |
+
|
160 |
+
## Processing Time
|
161 |
+
Please note that processing will take around 2 minutes.
|
162 |
+
""")
|
163 |
with gr.Tabs():
|
164 |
with gr.TabItem("🏆 Leaderboard"):
|
165 |
leaderboard_html = gr.HTML(create_html_table(format_leaderboard_df(load_leaderboard_data())))
|
requirements.txt
CHANGED
@@ -5,7 +5,7 @@ torchvision==0.15.2
|
|
5 |
transformers==4.46.3
|
6 |
tokenizers>=0.20,<0.21
|
7 |
safetensors>=0.4.1
|
8 |
-
evaluate==0.4.
|
9 |
gradio==5.7.1
|
10 |
huggingface-hub==0.25.1
|
11 |
panphon==0.21
|
|
|
5 |
transformers==4.46.3
|
6 |
tokenizers>=0.20,<0.21
|
7 |
safetensors>=0.4.1
|
8 |
+
evaluate==0.4.3
|
9 |
gradio==5.7.1
|
10 |
huggingface-hub==0.25.1
|
11 |
panphon==0.21
|