up

app.py

@@ -80,4 +80,36 @@ COLUMN_NAMES = {"librispeech-clean": "ls-clean", "librispeech-other": "ls-other"
 table = all_results.round(4)
 table = table.rename(columns=COLUMN_NAMES)
 
+# Streamlit
+st.markdown("# ESC: A Benchmark For Multi-Domain End-to-End Speech Recognition")
+
+st.markdown(
+    f"""
+    This is the leaderboard of the End-to end Speech Challenge (ESC).
+    Submitted systems are ranked by the **ESC Score** which is the average of 
+    all non-optional datasets: {', '.join(COLUMN_NAMES.values())}."""
+)
+
 st.table(table)
+
+#    *Sanchit Gandhi, Patrick Von Platen, and, Alexander M Rush*
+st.markdown(
+    """
+    ESC was proposed in *ESC: A Benchmark For Multi-Domain End-to-End Speech Recognition* by ...
+    \n 
+    The abstract of the paper is as follows: 
+    \n 
+    *Speech recognition applications cover a range of different audio and text distributions, with different speaking styles, background noise, transcription punctuation and character casing. However, many speech recognition systems require dataset-specific tuning (audio filtering, punctuation removal and normalisation of casing), therefore assuming a-priori knowledge of both the audio and text distributions. This tuning requirement can lead to systems failing to generalise to other datasets and domains. To promote the development of multi-domain speech systems, we introduce the End-to end Speech Challenge (ESC) for evaluating the performance of a single automatic speech recognition (ASR) system across a broad set of speech datasets. Benchmarked systems must use the same data pre- and post-processing algorithm across datasets - assuming the audio and text data distributions are a-priori unknown. We compare a series of state-of-the-art (SoTA) end-to-end (E2E) systems on this benchmark, demonstrating how a single speechsystem can be applied and evaluated on a wide range of data distributions. We find E2E systems to be effective across datasets: in a fair comparison, E2E systems achieve within 2.6% of SoTA systems tuned to a specific dataset. Our analysis reveals that transcription artefacts, such as punctuation and casing, pose difficulties for ASR systems and should be included in evaluation. We believe E2E benchmarking over a range of datasets promotes the research of multi-domain speech recognition systems.*
+    \n
+    For more information, please see the official submission on [OpenReview.net](https://openreview.net/forum?id=9OL2fIfDLK).
+    """
+)
+
+st.markdown("To submit to ESC, please click on the instructions below ↓")
+
+st.markdown("TODO: Add instructions ...")
+
+uploaded_file = st.file_uploader("Choose a file")
+
+if st.button('Submit'):
+    st.write('Computing scores ...')

references

@@ -0,0 +1 @@
+Subproject commit 0fb7cf24ffe299a088275d73542484efad3c2667

submissions

@@ -0,0 +1 @@
+Subproject commit dddf172b2fe07941ccb2d3c5ae60ba81806c7a3b