Update src/streamlit_app.py
Browse files- src/streamlit_app.py +15 -28
src/streamlit_app.py
CHANGED
|
@@ -192,32 +192,19 @@ with tab5:
|
|
| 192 |
We will build a test set that can be used for benchmarking ASR models in some of the 30 most spoken African languages. The benchmark dataset will be structured to consist of unique MP3 files and corresponding text files. We will ensure as much as possible that the benchmark datasets are as diverse as possible with dataset characteristics like gender, age, accent, variant, vocabulary, acoustic characteristics to help improve the accuracy of speech recognition models. The speech benchmark dataset will be reviewed, deemed highly quality, and split into dev, test and train sets. Due to the largely acoustic nature of African languages (mostly tonal, diacritical, etc.), a careful speech analysis of African languages is necessary and the benchmark dataset is important to spur more research in the African context.
|
| 193 |
|
| 194 |
""")
|
| 195 |
-
|
| 196 |
# Citation
|
| 197 |
-
|
| 198 |
-
|
| 199 |
-
|
| 200 |
-
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
|
| 204 |
-
|
| 205 |
-
|
| 206 |
-
|
| 207 |
-
|
| 208 |
-
|
| 209 |
-
|
| 210 |
-
|
| 211 |
-
|
| 212 |
-
"""
|
| 213 |
-
<script>
|
| 214 |
-
function copyText() {
|
| 215 |
-
const text = document.querySelector('textarea').value;
|
| 216 |
-
navigator.clipboard.writeText(text);
|
| 217 |
-
alert("Citation copied to clipboard!");
|
| 218 |
-
}
|
| 219 |
-
</script>
|
| 220 |
-
<button onclick="copyText()">Copy Citation</button>
|
| 221 |
-
""",
|
| 222 |
-
unsafe_allow_html=True
|
| 223 |
-
)
|
|
|
|
| 192 |
We will build a test set that can be used for benchmarking ASR models in some of the 30 most spoken African languages. The benchmark dataset will be structured to consist of unique MP3 files and corresponding text files. We will ensure as much as possible that the benchmark datasets are as diverse as possible with dataset characteristics like gender, age, accent, variant, vocabulary, acoustic characteristics to help improve the accuracy of speech recognition models. The speech benchmark dataset will be reviewed, deemed highly quality, and split into dev, test and train sets. Due to the largely acoustic nature of African languages (mostly tonal, diacritical, etc.), a careful speech analysis of African languages is necessary and the benchmark dataset is important to spur more research in the African context.
|
| 193 |
|
| 194 |
""")
|
|
|
|
| 195 |
# Citation
|
| 196 |
+
CITATION_TEXT = """@misc{asr-africa-2025,
|
| 197 |
+
title = {Automatic Speech Recognition for African Languages},
|
| 198 |
+
author = {Dr Joyce Nakatumba-Nabende, Dr Peter Nabende, Dr Andrew Katumba, Alvin Nahabwe},
|
| 199 |
+
year = 2025,
|
| 200 |
+
publisher = {Hugging Face},
|
| 201 |
+
howpublished = "\\url{https://huggingface.co/spaces/asr-africa/Automatic_Speech_Recognition_for_African_Languages}"
|
| 202 |
+
}"""
|
| 203 |
+
|
| 204 |
+
with st.expander("π Citation", expanded=False):
|
| 205 |
+
st.text_area(
|
| 206 |
+
"BibTeX snippet to cite this source",
|
| 207 |
+
value=CITATION_TEXT,
|
| 208 |
+
height=150,
|
| 209 |
+
disabled=True
|
| 210 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|