Update README.md

README.md

@@ -15,12 +15,13 @@ Configs:
 - max=10s: combine consecutive samples for a period shorter than 10 seconds
 - max: combine all the samples of a TEDx talk
 
-```
 
-dependencies (for much faster audio decoding):
+dependencies (only needed for much faster audio decoding):
  - ffmpeg: apt install ffmpeg
  - ffmpeg-python: pip install ffmpeg-python
 
+
+```
 @inproceedings{salesky2021mtedx,
   title={Multilingual TEDx Corpus for Speech Recognition and Translation},
   author={Elizabeth Salesky and Matthew Wiesner and Jacob Bremerman and Roldano Cattoni and Matteo Negri and Marco Turchi and Douglas W. Oard and Matt Post},