Update app.py

app.py

@@ -36,6 +36,7 @@ The Open Universal Arabic ASR Leaderboard will be a continuous benchmark project
 | [MASC (Clean-Test)](https://ieee-dataport.org/open-access/masc-massive-arabic-speech-corpus)    | 7              | 10.5        |
 | [MASC (Noisy-Test)](https://ieee-dataport.org/open-access/masc-massive-arabic-speech-corpus)    | 8              | 14.9        |
 | [MGB-2](http://www.mgb-challenge.org/MGB-2.html)                                                | Unspecified    | 9.6         |
+| [Casablanca](https://huggingface.co/datasets/UBC-NLP/Casablanca)                                | 8              | 7.7         |
 
 ## In-depth Analysis
 We also provide a comprehensive analysis of the models' robustness, speaker adaptation, inference efficiency and memory consumption.
@@ -46,29 +47,31 @@ We also provide a comprehensive analysis of the models' robustness, speaker adap
 def styled_message(message):
     return f"<p style='color: green; font-size: 20px; text-align: center;'>{message}</p>"
 
-LAST_UPDATED = "NOV 6th 2024"
+LAST_UPDATED = "Jan 12th 2025:\nNew models included: nvidia-Parakeet-1.1b-concat and nvidia-Parakeet-1.1b-universal\nNew dataset included: Casablanca"
 
 
 results = {
-    "Model": ["nvidia-conformer-large-arabic (lm)", "nvidia-conformer-large-arabic (greedy)", "openai/whisper-large-v3", "facebook/seamless-m4t-v2-large", "openai/whisper-large-v3-turbo", "openai/whisper-large-v2", "openai/whisper-large", "asafaya/hubert-large-arabic-transcribe/", "openai/whisper-medium", "facebook/mms-1b-all", "openai/whisper-small", "whitefox123/w2v-bert-2.0-arabic-4", "jonatasgrosman/wav2vec2-large-xlsr-53-arabic", "speechbrain/asr-wav2vec2-commonvoice-14-ar"],
-    "Average WER⬇️": [25.71, 27.46, 29.87, 32.55, 33.30, 34.04, 36.65, 39.29, 39.60, 47.86, 48.62, 52.18, 54.63, 60.15],
-    "Average CER": [10.02, 9.94, 13.65, 14.47, 15.68, 16.26, 17.44, 13.61, 19.10, 17.66, 16.79, 25.15, 21.46, 26.64],
-    "SADA WER": [44.52, 47.26, 55.96, 62.52, 60.36, 57.46, 63.24, 67.82, 67.71, 77.48, 78.02, 87.34, 86.82, 88.54],
-    "SADA CER": [23.76, 22.54, 34.62, 37.61, 37.67, 36.59, 40.16, 31.83, 43.83, 37.50, 33.17, 56.75, 44.20, 50.28],
-    "Common Voice WER": [8.80, 10.60, 17.83, 21.70, 25.73, 21.77, 26.04, 8.01, 28.07, 26.52, 24.18, 41.79, 23.00, 29.17],
-    "Common Voice CER": [2.77, 3.05, 5.74, 6.24, 10.89, 7.44, 9.61, 2.37, 10.38, 7.21, 6.79, 15.75, 6.64, 9.85],
-    "MASC(clean-test) WER": [23.74, 24.12, 24.66, 25.04, 25.51, 27.25, 28.89, 32.94, 29.99, 38.82, 35.93, 37.82, 42.75, 49.10],
-    "MASC(clean-test) CER": [5.63, 5.63, 7.24, 7.19, 7.55, 8.28, 9.05, 7.15, 8.98, 10.36, 9.01, 11.92, 11.87, 16.37],
-    "MASC(noisy-test) WER": [34.29, 35.64, 34.63, 33.24, 37.16, 38.55, 40.79, 50.16, 42.91, 57.33, 56.36, 53.28, 64.27, 69.57],
-    "MASC(noisy-test) CER": [11.07, 11.02, 12.89, 11.92, 13.93, 15.49, 16.31, 15.62, 17.49, 19.76, 19.43, 21.93, 24.17, 30.17],
-    "MGB-2 WER": [17.20, 19.69, 16.26, 20.23, 17.75, 25.17, 24.28, 37.51, 29.32, 39.16, 48.64, 40.66, 56.29, 64.37],
-    "MGB-2 CER": [6.87, 7.46, 7.74, 9.37, 8.34, 13.48, 12.10, 11.07, 14.82, 13.48, 15.56, 19.39, 20.44, 26.56],
+    "Model": ["nvidia-conformer-large-arabic (lm)", "nvidia-conformer-large-arabic (greedy)", "openai/whisper-large-v3", "facebook/seamless-m4t-v2-large", "openai/whisper-large-v3-turbo", "openai/whisper-large-v2", "openai/whisper-large", "asafaya/hubert-large-arabic-transcribe/", "openai/whisper-medium", "nvidia-Parakeet-1.1b-concat", "nvidia-Parakeet-1.1b-universal", "facebook/mms-1b-all", "openai/whisper-small", "whitefox123/w2v-bert-2.0-arabic-4", "jonatasgrosman/wav2vec2-large-xlsr-53-arabic", "speechbrain/asr-wav2vec2-commonvoice-14-ar"],
+    "Average WER⬇️": [32.91, 34.74, 36.86, 38.16, 40.05, 40.20, 42.57, 45.50, 45.57, 46.54, 51.96, 54.54, 55.13, 58.13, 60.98, 65.74],
+    "Average CER": [13.84, 13.37, 17.21, 17.03, 18.87, 19.55, 20.49, 17.35, 22.27, 23.88, 25.19, 21.45, 21.68, 27.62, 25.61, 30.93],
+    "SADA WER": [44.52, 47.26, 55.96, 62.52, 60.36, 57.46, 63.24, 67.82, 67.71, 70.70, 73.58, 77.48, 78.02, 87.34, 86.82, 88.54],
+    "SADA CER": [23.76, 22.54, 34.62, 37.61, 37.67, 36.59, 40.16, 31.83, 43.83, 46.70, 49.48, 37.50, 33.17, 56.75, 44.20, 50.28],
+    "Common Voice WER": [8.80, 10.60, 17.83, 21.70, 25.73, 21.77, 26.04, 8.01, 28.07, 26.34, 40.01, 26.52, 24.18, 41.79, 23.00, 29.17],
+    "Common Voice CER": [2.77, 3.05, 5.74, 6.24, 10.89, 7.44, 9.61, 2.37, 10.38, 9.82, 14.64, 7.21, 6.79, 15.75, 6.64, 9.85],
+    "MASC(clean-test) WER": [23.74, 24.12, 24.66, 25.04, 25.51, 27.25, 28.89, 32.94, 29.99, 30.49, 36.16, 38.82, 35.93, 37.82, 42.75, 49.10],
+    "MASC(clean-test) CER": [5.63, 5.63, 7.24, 7.19, 7.55, 8.28, 9.05, 7.15, 8.98, 8.41, 10.29, 10.36, 9.01, 11.92, 11.87, 16.37],
+    "MASC(noisy-test) WER": [34.29, 35.64, 34.63, 33.24, 37.16, 38.55, 40.79, 50.16, 42.91, 45.95, 50.03, 57.33, 56.36, 53.28, 64.27, 69.57],
+    "MASC(noisy-test) CER": [11.07, 11.02, 12.89, 11.92, 13.93, 15.49, 16.31, 15.62, 17.49, 18.72, 20.09, 19.76, 19.43, 21.93, 24.17, 30.17],
+    "MGB-2 WER": [17.20, 19.69, 16.26, 20.23, 17.75, 25.17, 24.28, 37.51, 29.32, 24.94, 30.68, 39.16, 48.64, 40.66, 56.29, 64.37],
+    "MGB-2 CER": [6.87, 7.46, 7.74, 9.37, 8.34, 13.48, 12.10, 11.07, 14.82, 9.87, 11.36, 13.48, 15.56, 19.39, 20.44, 26.56],
+    "Casablanca WER": [68.90, 71.13, 71.81, 66.25, 73.79, 71.01, 72.18, 76.53, 75.44, 80.80, 81.30, 87.95, 87.64, 87.88, 92.72, 93.68],
+    "Casablanca CER": [32.97, 30.50, 35.04, 29.85, 34.83, 36.00, 35.71, 36.03, 38.12, 49.77, 45.31, 40.41, 46.12, 39.99, 46.33, 52.36],
 }
 
 original_df = pd.DataFrame(results)
 original_df.sort_values(by="Average WER⬇️", inplace=True)
 
-TYPES = ['str', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number']
+TYPES = ['str', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number', 'number']
 
 
 def request_model(model_text):