simplified app

app.py

@@ -1,150 +1,50 @@
-# original app: https://huggingface.co/spaces/xianbao/whisper-v3-zero
-
-import torch
-
 import gradio as gr
+from transformers import WhisperProcessor, WhisperForConditionalGeneration
+import torch
+import librosa
 import spaces
-import yt_dlp as youtube_dl
-from transformers import pipeline
-from transformers.pipelines.audio_utils import ffmpeg_read
-
-import tempfile
-import os
-
-MODEL_NAME = "TheirStory/whisper-small-xhosa"
-BATCH_SIZE = 8
-FILE_LIMIT_MB = 1000
-YT_LENGTH_LIMIT_S = 3600  # limit to 1 hour YouTube files
 
-device = 0 if torch.cuda.is_available() else "cpu"
-
-pipe = pipeline(
-    task="automatic-speech-recognition",
-    model=MODEL_NAME,
-    chunk_length_s=30,
-    device=device,
-)
+# Load the model and processor
+model_name = "TheirStory/whisper-small-xhosa"
+processor = WhisperProcessor.from_pretrained(model_name)
+model = WhisperForConditionalGeneration.from_pretrained(model_name)
 
 @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.")
-
-    text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
-    return  text
-
-
-def _return_yt_html_embed(yt_url):
-    video_id = yt_url.split("?v=")[-1]
-    HTML_str = (
-        f'<center> <iframe width="500" height="320" src="https://www.youtube.com/embed/{video_id}"> </iframe>'
-        " </center>"
-    )
-    return HTML_str
-
-def download_yt_audio(yt_url, filename):
-    info_loader = youtube_dl.YoutubeDL()
-    
-    try:
-        info = info_loader.extract_info(yt_url, download=False)
-    except youtube_dl.utils.DownloadError as err:
-        raise gr.Error(str(err))
-    
-    file_length = info["duration_string"]
-    file_h_m_s = file_length.split(":")
-    file_h_m_s = [int(sub_length) for sub_length in file_h_m_s]
-    
-    if len(file_h_m_s) == 1:
-        file_h_m_s.insert(0, 0)
-    if len(file_h_m_s) == 2:
-        file_h_m_s.insert(0, 0)
-    file_length_s = file_h_m_s[0] * 3600 + file_h_m_s[1] * 60 + file_h_m_s[2]
-    
-    if file_length_s > YT_LENGTH_LIMIT_S:
-        yt_length_limit_hms = time.strftime("%HH:%MM:%SS", time.gmtime(YT_LENGTH_LIMIT_S))
-        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"}
-    
-    with youtube_dl.YoutubeDL(ydl_opts) as ydl:
-        try:
-            ydl.download([yt_url])
-        except youtube_dl.utils.ExtractorError as err:
-            raise gr.Error(str(err))
+def transcribe_audio(audio):
+    if torch.cuda.is_available():
+        model = model.to("cuda")
 
+    # Load the audio file
+    if isinstance(audio, str):  # If it's a file path
+        audio_array, sampling_rate = librosa.load(audio, sr=16000)
+    else:  # If it's a tuple (audio_array, sampling_rate)
+        audio_array, sampling_rate = audio
 
-def yt_transcribe(yt_url, task, max_filesize=75.0):
-    html_embed_str = _return_yt_html_embed(yt_url)
+    # Process the audio
+    input_features = processor(audio_array, sampling_rate=sampling_rate, return_tensors="pt").input_features
 
-    with tempfile.TemporaryDirectory() as tmpdirname:
-        filepath = os.path.join(tmpdirname, "video.mp4")
-        download_yt_audio(yt_url, filepath)
-        with open(filepath, "rb") as f:
-            inputs = f.read()
+    if torch.cuda.is_available():
+        input_features = input_features.to("cuda")
 
-    inputs = ffmpeg_read(inputs, pipe.feature_extractor.sampling_rate)
-    inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate}
+    # Generate token ids
+    generated_ids = model.generate(input_features)
 
-    text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
+    # Decode token ids to text
+    transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
 
-    return html_embed_str, text
+    return transcription
 
-
-demo = gr.Blocks()
-
-mf_transcribe = gr.Interface(
-    fn=transcribe,
-    inputs=[
-        gr.Audio(type="filepath"),
-        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
-    ],
-    outputs="text",
-    theme="huggingface",
-    title="Whisper Large V3: Transcribe Audio",
-    description=(
-        "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the OpenAI Whisper"
-        f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
-        " of arbitrary length."
-    ),
-    allow_flagging="never",
-)
-
-file_transcribe = gr.Interface(
-    fn=transcribe,
+# Create the Gradio interface
+iface = gr.Interface(
+    fn=transcribe_audio,
     inputs=[
-        gr.Audio(type="filepath", label="Audio file"),
-        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
+        gr.Audio(source="microphone", type="numpy", label="Record Audio"),
+        gr.Audio(source="upload", type="filepath", label="Upload Audio File")
     ],
     outputs="text",
-    theme="huggingface",
-    title="Whisper Large V3: Transcribe Audio",
-    description=(
-        "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the OpenAI Whisper"
-        f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
-        " of arbitrary length."
-    ),
-    allow_flagging="never",
-)
-
-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", "translate"], label="Task", value="transcribe")
-    ],
-    outputs=["html", "text"],
-    theme="huggingface",
-    title="Whisper Large V3: Transcribe YouTube",
-    description=(
-        "Transcribe long-form YouTube videos with the click of a button! Demo uses the OpenAI Whisper checkpoint"
-        f" [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe video files of"
-        " arbitrary length."
-    ),
-    allow_flagging="never",
+    title="Xhosa Audio Transcription",
+    description="Record or upload Xhosa audio to get its transcription using the TheirStory/whisper-small-xhosa model."
 )
 
-with demo:
-    gr.TabbedInterface([mf_transcribe, file_transcribe, yt_transcribe], ["Microphone", "Audio file", "YouTube"])
-
-demo.launch()
+# Launch the app
+iface.launch()

requirements.txt

@@ -1,3 +1,4 @@
 git+https://github.com/huggingface/transformers
 torch
-yt-dlp
+yt-dlp
+librosa