hotfix 2.0

README.md

@@ -1,13 +1,11 @@
 ---
 title: Verbalens
 emoji: 📚
-colorFrom: purple
-colorTo: purple
+colorFrom: indigo
+colorTo: red
 sdk: gradio
 sdk_version: 5.20.0
 app_file: app.py
 pinned: false
 short_description: this is a first prototype of verbalens using whisper and nem
 ---
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -1,33 +1,36 @@
 import spaces
 import torch
-from faster_whisper import WhisperModel
 
 import gradio as gr
 import yt_dlp as youtube_dl
-
+from transformers import pipeline
 from transformers.pipelines.audio_utils import ffmpeg_read
 
 import tempfile
 import os
 
-MODEL_NAME = "large-v3"
+MODEL_NAME = "openai/whisper-large-v3-turbo"
 BATCH_SIZE = 8
 FILE_LIMIT_MB = 1000
-YT_LENGTH_LIMIT_S = 600  # limit to 1 hour YouTube files
+YT_LENGTH_LIMIT_S = 3600  # limit to 1 hour YouTube files
 
 device = 0 if torch.cuda.is_available() else "cpu"
 
-model = WhisperModel(MODEL_NAME, device=device, compute_type="float16" if device == "cuda" else "int8")
+pipe = pipeline(
+    task="automatic-speech-recognition",
+    model=MODEL_NAME,
+    chunk_length_s=30,
+    device=device,
+)
+
 
 @spaces.GPU
 def transcribe(inputs, task):
     if inputs is None:
         raise gr.Error("No audio file submitted! Please upload or record an audio file before submitting your request.")
-    segments, info = model.transcribe(input, beam_size=5, vad_filter=True, word_timestamps=False)
-    transcription = " ".join([segment.text for segment in segments])
-    
-    #text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
-    return  transcription
+
+    text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
+    return  text
 
 
 def _return_yt_html_embed(yt_url):
@@ -61,10 +64,7 @@ def download_yt_audio(yt_url, filename):
         file_length_hms = time.strftime("%HH:%MM:%SS", time.gmtime(file_length_s))
         raise gr.Error(f"Maximum YouTube length is {yt_length_limit_hms}, got {file_length_hms} YouTube video.")
     
-    ydl_opts = {"outtmpl": filename, "format": "worstvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best",'postprocessors': [{
-            'key': 'FFmpegExtractAudio',
-            'preferredcodec': 'mp3',
-        }]}
+    ydl_opts = {"outtmpl": filename, "format": "worstvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best"}
     
     with youtube_dl.YoutubeDL(ydl_opts) as ydl:
         try:
@@ -77,32 +77,32 @@ def yt_transcribe(yt_url, task, max_filesize=75.0):
     html_embed_str = _return_yt_html_embed(yt_url)
 
     with tempfile.TemporaryDirectory() as tmpdirname:
-        filepath = os.path.join(tmpdirname, "audio.mp3")
+        filepath = os.path.join(tmpdirname, "video.mp4")
         download_yt_audio(yt_url, filepath)
+        with open(filepath, "rb") as f:
+            inputs = f.read()
 
-    #inputs = ffmpeg_read(inputs, pipe.feature_extractor.sampling_rate)
-    #inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate}
+    inputs = ffmpeg_read(inputs, pipe.feature_extractor.sampling_rate)
+    inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate}
 
-    #text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
-    segments, info = model.transcribe(filepath, beam_size=5, vad_filter=True, word_timestamps=False)
-    transcription = " ".join([segment.text for segment in segments])
+    text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
 
-    return html_embed_str, transcription
+    return html_embed_str, text
 
 
-demo = gr.Blocks()
+demo = gr.Blocks(theme=gr.themes.Ocean())
 
 mf_transcribe = gr.Interface(
     fn=transcribe,
     inputs=[
         gr.Audio(sources="microphone", type="filepath"),
-        gr.Radio(["transcribe", "analysis"], label="Task", value="transcribe"),
+        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
     ],
     outputs="text",
-    title="VerbaLens Project: Demo 1",
+    title="Whisper Large V3 Turbo: Transcribe Audio",
     description=(
         "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the"
-        f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}), Nemo diarization and Transformers to transcribe audio files"
+        f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
         " of arbitrary length."
     ),
     allow_flagging="never",
@@ -115,11 +115,11 @@ file_transcribe = gr.Interface(
         gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
     ],
     outputs="text",
-    title="Verbalens Project: Demo 1 prototype",
+    title="Whisper Large V3: Transcribe Audio",
     description=(
-      "Transcribe long-form videos with the click of a button! Demo uses the checkpoint"
-        f" [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}), Nemo Diarization and Transformers to transcribe video files of"
-        " arbitrary length."
+        "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the"
+        f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
+        " of arbitrary length."
     ),
     allow_flagging="never",
 )
@@ -128,12 +128,14 @@ yt_transcribe = gr.Interface(
     fn=yt_transcribe,
     inputs=[
         gr.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL"),
-        gr.Radio(["transcribe", "analysis"], label="Task", value="transcribe")
+        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe")
     ],
     outputs=["html", "text"],
-    title="Verbalens Project: Demo 1 prototype",
+    title="Whisper Large V3: Transcribe YouTube",
     description=(
-        "Transcribe long-form videos with the click of a button! Demo uses the checkpoint Nemo Diarization and Transformers to transcribe video files of arbitrary length."
+        "Transcribe long-form YouTube videos with the click of a button! Demo uses the checkpoint"
+        f" [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe video files of"
+        " arbitrary length."
     ),
     allow_flagging="never",
 )
@@ -141,4 +143,4 @@ yt_transcribe = gr.Interface(
 with demo:
     gr.TabbedInterface([mf_transcribe, file_transcribe, yt_transcribe], ["Microphone", "Audio file", "YouTube"])
 
-demo.queue().launch()
+demo.queue().launch(ssr_mode=False)