Update app.py

app.py

@@ -1,151 +1,99 @@
 import torch
-
 import gradio as gr
+from faster_whisper import WhisperModel
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
+from pydub import AudioSegment
 import yt_dlp as youtube_dl
-from transformers import pipeline
-from transformers.pipelines.audio_utils import ffmpeg_read
-
 import tempfile
+from transformers.pipelines.audio_utils import ffmpeg_read
+from gradio.components import Audio, Dropdown, Radio, Textbox
 import os
-
-MODEL_NAME = "openai/whisper-large-v2"
-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,
-)
+os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
 
-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
-
+# Paramètres
+FILE_LIMIT_MB = 1000
+YT_LENGTH_LIMIT_S = 3600  # Limite de 1 heure pour les vidéos YouTube
+
+# Charger les codes de langue
+from flores200_codes import flores_codes
+lang_codes = list(flores_codes.keys())
+
+# Fonction pour déterminer le device
+def set_device():
+    return torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+device = set_device()
+
+# Charger les modèles une seule fois
+model_dict = {}
+def load_models():
+    global model_dict
+    if not model_dict:
+        model_name_dict = {
+            #'nllb-distilled-1.3B': 'facebook/nllb-200-distilled-1.3B',
+            'nllb-distilled-600M': 'facebook/nllb-200-distilled-600M',
+            #'nllb-1.3B': 'facebook/nllb-200-1.3B',
+            #'nllb-distilled-1.3B': 'facebook/nllb-200-distilled-1.3B',
+            #'nllb-3.3B': 'facebook/nllb-200-3.3B',
+            # 'nllb-distilled-600M': 'facebook/nllb-200-distilled-600M',
+        }
+        for call_name, real_name in model_name_dict.items():
+            model = AutoModelForSeq2SeqLM.from_pretrained(real_name)
+            tokenizer = AutoTokenizer.from_pretrained(real_name)
+            model_dict[call_name+'_model'] = model
+            model_dict[call_name+'_tokenizer'] = tokenizer
+
+load_models()
+
+# Fonction pour la transcription
+def transcribe_audio(audio_file):
+    model_size = "large-v2"
+    model = WhisperModel(model_size)
+    # model = WhisperModel(model_size, device=device, compute_type="int8")
+    segments, _ = model.transcribe(audio_file, beam_size=1)
+    transcriptions = [("[%.2fs -> %.2fs]" % (seg.start, seg.end), seg.text) for seg in segments]
+    return transcriptions
+
+# Fonction pour la traduction
+def traduction(text, source_lang, target_lang):
+    model_name = "nllb-distilled-600M"
+    model = model_dict[model_name + "_model"]
+    tokenizer = model_dict[model_name + "_tokenizer"]
+    translator = pipeline("translation", model=model, tokenizer=tokenizer)
+    return translator(text, src_lang=flores_codes[source_lang], tgt_lang=flores_codes[target_lang])[0]["translation_text"]
+
+# Fonction principale
+def full_transcription_and_translation(audio_file, source_lang, target_lang):
+    transcriptions = transcribe_audio(audio_file)
+    translations = [(timestamp, traduction(text, source_lang, target_lang)) for timestamp, text in transcriptions]
+    return transcriptions, translations
+
+# Téléchargement audio YouTube
 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.")
+    with youtube_dl.YoutubeDL({'format': 'bestaudio'}) as ydl:
+        ydl.download([yt_url])
 
-    ydl_opts = {"outtmpl": filename, "format": "worstvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best"}
+# Interface Gradio
+def gradio_interface(audio_file, source_lang, target_lang):
+    transcriptions, translations = full_transcription_and_translation(audio_file, source_lang, target_lang)
+    transcribed_text = '\n'.join([f"{timestamp}: {text}" for timestamp, text in transcriptions])
+    translated_text = '\n'.join([f"{timestamp}: {text}" for timestamp, text in translations])
+    return transcribed_text, translated_text
 
-    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 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, "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}
-
-    text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
-
-    return html_embed_str, text
-
-
-demo = gr.Blocks()
-
-mf_transcribe = gr.Interface(
-    fn=transcribe,
+iface = gr.Interface(
+    fn=gradio_interface,
     inputs=[
-        gr.inputs.Audio(source="microphone", type="filepath", optional=True),
-        gr.inputs.Radio(["transcribe", "translate"], label="Task", default="transcribe"),
+        gr.Audio(type="filepath"),
+        gr.Dropdown(lang_codes, value='French', label='Source Language'),
+        gr.Dropdown(lang_codes, value='English', label='Target Language'),
     ],
-    outputs="text",
-    layout="horizontal",
-    theme="huggingface",
-    title="Whisper Large V2: 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}) and 🤗 Transformers to transcribe audio files"
-        " of arbitrary length."
-    ),
-    allow_flagging="never",
+    outputs=[
+        gr.Textbox(label="Transcribed Text"),
+        gr.Textbox(label="Translated Text")
+    ]
 )
 
-file_transcribe = gr.Interface(
-    fn=transcribe,
-    inputs=[
-        gr.inputs.Audio(source="upload", type="filepath", optional=True, label="Audio file"),
-        gr.inputs.Radio(["transcribe", "translate"], label="Task", default="transcribe"),
-    ],
-    outputs="text",
-    layout="horizontal",
-    theme="huggingface",
-    title="Whisper Large V2: 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}) and 🤗 Transformers to transcribe audio files"
-        " of arbitrary length."
-    ),
-    allow_flagging="never",
-)
-
-yt_transcribe = gr.Interface(
-    fn=yt_transcribe,
-    inputs=[
-        gr.inputs.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL"),
-        gr.inputs.Radio(["transcribe", "translate"], label="Task", default="transcribe")
-    ],
-    outputs=["html", "text"],
-    layout="horizontal",
-    theme="huggingface",
-    title="Whisper Large V2: Transcribe YouTube",
-    description=(
-        "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",
-)
-
-with demo:
-    gr.TabbedInterface([mf_transcribe, file_transcribe, yt_transcribe], ["Microphone", "Audio file", "YouTube"])
-
-demo.launch(enable_queue=True)
+iface.launch()