add spinners

app.py

@@ -166,7 +166,8 @@ def strike(url, speedup_factor, min_speedup, max_speedup, max_num_segments):
 
     min_speedup = max(0.5, min_speedup)  # ffmpeg limit
 
-    name = download(url, YDL_OPTS)
+    with st.spinner("downloading..."):
+        name = download(url, YDL_OPTS)
     assert name.endswith(".m4a")
     name = name.split(".m4a")[0].split("/")[-1]
 
@@ -174,13 +175,15 @@ def strike(url, speedup_factor, min_speedup, max_speedup, max_num_segments):
     transcript_path = os.path.join(DATA_DIR, "%s.json" % name)
     output_path = os.path.join(DATA_DIR, "%s_smooth.mp3" % name)
 
-    segments = transcribe(audio_path, transcript_path)
+    with st.spinner("transcribing..."):
+        segments = transcribe(audio_path, transcript_path)
 
     seg_durations = compute_seg_durations(segments)
 
-    info_densities = compute_info_densities(
-        segments, seg_durations, llm, tokenizer, device
-    )
+    with st.spinner("calculating information density..."):
+        info_densities = compute_info_densities(
+            segments, seg_durations, llm, tokenizer, device
+        )
 
     total_duration = segments[-1]["end"] - segments[0]["start"]
     min_sec_leaf = total_duration / max_num_segments
@@ -201,7 +204,8 @@ def strike(url, speedup_factor, min_speedup, max_speedup, max_num_segments):
         total_duration,
     )
 
-    cat_clips(squashed_times, speedups, audio_path, output_path)
+    with st.spinner("stitching segments..."):
+        cat_clips(squashed_times, speedups, audio_path, output_path)
 
     spedup_total_duration, actual_speedup_factor = compute_actual_speedup(
         squashed_durations, speedups, total_duration
@@ -216,19 +220,24 @@ def strike(url, speedup_factor, min_speedup, max_speedup, max_num_segments):
     data = [times, info_densities / np.log(2), annotations]
     cols = ["time (minutes)", "bits per second", "transcript"]
     df = pd.DataFrame(list(zip(*data)), columns=cols)
-
+    min_time = segments[0]["start"] / 60
+    max_time = segments[-1]["end"] / 60
     lines = (
         alt.Chart(df, title="information rate")
         .mark_line(color="gray", opacity=0.5)
         .encode(
-            x=cols[0],
+            x=alt.X(cols[0], scale=alt.Scale(domain=(min_time, max_time))),
             y=cols[1],
         )
     )
     dots = (
         alt.Chart(df)
         .mark_circle(size=50, opacity=1)
-        .encode(x=cols[0], y=cols[1], tooltip=["transcript"])
+        .encode(
+            x=alt.X(cols[0], scale=alt.Scale(domain=(min_time, max_time))),
+            y=cols[1],
+            tooltip=["transcript"],
+        )
     )
     st.altair_chart((lines + dots).interactive(), use_container_width=True)
 
@@ -245,7 +254,7 @@ def strike(url, speedup_factor, min_speedup, max_speedup, max_num_segments):
         alt.Chart(df, title="speedup based on information rate")
         .mark_line()
         .encode(
-            x=cols[0],
+            x=alt.X(cols[0], scale=alt.Scale(domain=(min_time, max_time))),
             y=alt.Y(
                 cols[1],
                 scale=alt.Scale(
@@ -274,9 +283,8 @@ with st.form("my_form"):
     if submitted:
         st.write("original video:")
         st.video(url)
-        with st.spinner("processing audio..."):
-            output_path = strike(
-                url, speedup_factor, min_speedup, max_speedup, max_num_segments
-            )
+        output_path = strike(
+            url, speedup_factor, min_speedup, max_speedup, max_num_segments
+        )
         st.write("processed audio:")
         st.audio(output_path)