Add progress bar and report overall processing time.

Dockerfile

 FROM ghcr.io/opennmt/ctranslate2:latest-ubuntu20.04-cuda11.2
 
 # install faster-whisper
-RUN python3 -m pip --no-cache-dir install faster-whisper
+RUN python3 -m pip --no-cache-dir install faster-whisper tqdm
 
 # preload model
 RUN python3 -c "import faster_whisper; faster_whisper.download_model('large-v2')"

main.py

 from faster_whisper import WhisperModel
-from datetime import timedelta
 from time import perf_counter
+from tqdm import tqdm
 
 model_size = "large-v2"
 
@@ -18,23 +18,30 @@ segments, info = model.transcribe("/input/audiofile", language="de", beam_size=5
 with open("/output/transcript.vtt", "w", encoding="utf-8") as f:
     f.write("WEBVTT\n\nNOTE This transcript was automatically generated.")
 
-    print("Processing...", end="")
-    for segment in segments:
-        start = timedelta(seconds=segment.start)
-        end = timedelta(seconds=segment.end)
-        text = segment.text[1:]
+    print("Start processing...")
+    with tqdm(total=info.duration, leave=False) as pbar:
+        previousEnd = 0
+        for segment in segments:
+            # calculate cue times
+            startM, startS = divmod(segment.start, 60)
+            startH, startM = divmod(startM, 60)
 
-        entry = "{} --> {}\n{}".format(start, end, text)
+            endM, endS = divmod(segment.end, 60)
+            endH, endM = divmod(endM, 60)
 
-        f.write("\n\n")
-        f.write(entry)
+            # write cue & text
+            f.write("\n\n{:02.0f}:{:02.0f}:{:06.3f} --> {:02.0f}:{:02.0f}:{:06.3f}\n".format(startH, startM, startS, endH, endM, endS))
+            f.write(segment.text[1:])
 
-        progress = segment.end / info.duration
-        print("{:2.0%}".format(progress if progress <= 1 else 1), end="...")
+            # update progressbar
+            pbar.update(segment.end - previousEnd)
+            previousEnd = segment.end
 
 print("Done!")
 
-tEnd = perf_counter()
-tDelta = tEnd - tStart
+tDelta = perf_counter() - tStart
+tDeltaM, tDeltaS = divmod(tDelta, 60)
 
-print("Processing time: {}s".format(timedelta(seconds=tDelta)))
+durationM, durationS = divmod(info.duration, 60)
+
+print("Processed {:02.0f}m {:02.0f}s audio in {:02.0f}m {:02.0f}s".format(durationM, durationS, tDeltaM, tDeltaS))