Migrate to use small-q5_1 for faster inference

Dockerfile

@@ -6,7 +6,7 @@ RUN mkdir /whisper && \
 
 WORKDIR /whisper
 
-ARG model=medium-q5_0
+ARG model=small-q5_1
 RUN bash ./models/download-ggml-model.sh "${model}"
 RUN make main
 
@@ -26,7 +26,7 @@ RUN pip install -r $HOME/root/requirements.txt
 COPY --chown=user main.py $HOME/root
 COPY --chown=user youtubeaudio.py $HOME/root
 
-ARG model=medium-q5_0
+ARG model=small-q5_1
 ENV model=$model
 RUN mkdir $HOME/root/models
 RUN mkdir -p -m 775 /tmp/holosubs/tracks /tmp/holosubs/videos