Upload 4 files

app.py

@@ -0,0 +1,146 @@
+import spaces
+import torch
+
+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 = "openai/whisper-large-v3-turbo"
+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,
+)
+
+
+@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))
+
+@spaces.GPU
+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(theme=gr.themes.Ocean())
+
+mf_transcribe = gr.Interface(
+    fn=transcribe,
+    inputs=[
+        gr.Audio(sources="microphone", type="filepath"),
+        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
+    ],
+    outputs="text",
+    title="清華大學多模態課程＆廖老師嫡傳弟子-第二組 「語音轉文字」model",
+    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",
+)
+
+file_transcribe = gr.Interface(
+    fn=transcribe,
+    inputs=[
+        gr.Audio(sources="upload", type="filepath", label="Audio file"),
+        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
+    ],
+    outputs="text",
+    title="Whisper Large V3: 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.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"],
+    title="Whisper Large V3: 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.queue().launch(ssr_mode=False)

packages.txt

@@ -0,0 +1 @@
+ffmpeg

requirements.txt

@@ -0,0 +1,3 @@
+transformers
+yt-dlp
+torch

whisper_notebook.ipynb

@@ -0,0 +1,192 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "provenance": [],
+      "gpuType": "T4"
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "language_info": {
+      "name": "python"
+    },
+    "accelerator": "GPU"
+  },
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "source": [
+        "# Whisper v3 is here!\n",
+        "\n",
+        "Whisper v3 is a new model open sourced by OpenAI. The model can do multilingual transcriptions and is quite impressive. For example, you can change from English to Spanish or Chinese in the middle of a sentence and it will work well!\n",
+        "\n",
+        "The model can be run in a free Google Colab instance and is integrated into `transformers` already, so switching can be a very smooth process if you already use the previous versions."
+      ],
+      "metadata": {
+        "id": "OXaUqiE-eyXM"
+      }
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "WFQeUT9EcIcK"
+      },
+      "outputs": [],
+      "source": [
+        "%%capture\n",
+        "!pip install git+https://github.com/huggingface/transformers gradio"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "Let's use the high level `pipeline` from the `transformers` library to load the model."
+      ],
+      "metadata": {
+        "id": "sZONes21fHTA"
+      }
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "import torch\n",
+        "from transformers import pipeline\n",
+        "\n",
+        "pipe = pipeline(\"automatic-speech-recognition\",\n",
+        "               \"openai/whisper-large-v3\",\n",
+        "               torch_dtype=torch.float16,\n",
+        "               device=\"cuda:0\")"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "DvBdwMdPcr-Y",
+        "outputId": "47f32218-fd85-49ea-d880-d31577bcf9b8"
+      },
+      "execution_count": null,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stderr",
+          "text": [
+            "Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n",
+            "Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "pipe(\"https://cdn-media.huggingface.co/speech_samples/sample1.flac\")"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "GZFkIyhjc0Nc",
+        "outputId": "f1463431-3e08-4438-815f-b71e5e7a1503"
+      },
+      "execution_count": null,
+      "outputs": [
+        {
+          "output_type": "execute_result",
+          "data": {
+            "text/plain": [
+              "{'text': \" going along slushy country roads and speaking to damp audiences in draughty schoolrooms day after day for a fortnight he'll have to put in an appearance at some place of worship on sunday morning and he can come to us immediately afterwards\"}"
+            ]
+          },
+          "metadata": {},
+          "execution_count": 2
+        }
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "Let's now build a quick Gradio demo where we can play with the model directly using our microphone! You can run this code in a Google Colab instance (or locally!) or just head to the <a href=\"https://huggingface.co/spaces/hf-audio/whisper-large-v3\" target=\"_blank\">Space</a> to play directly with it online."
+      ],
+      "metadata": {
+        "id": "pt3YtM_PfTQY"
+      }
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "import gradio as gr\n",
+        "\n",
+        "def transcribe(inputs):\n",
+        "    if inputs is None:\n",
+        "        raise gr.Error(\"No audio file submitted! Please record an audio before submitting your request.\")\n",
+        "\n",
+        "    text = pipe(inputs, generate_kwargs={\"task\": \"transcribe\"}, return_timestamps=True)[\"text\"]\n",
+        "    return text\n",
+        "\n",
+        "demo = gr.Interface(\n",
+        "    fn=transcribe,\n",
+        "    inputs=[\n",
+        "        gr.Audio(sources=[\"microphone\", \"upload\"], type=\"filepath\"),\n",
+        "    ],\n",
+        "    outputs=\"text\",\n",
+        "    title=\"Whisper Large V3: Transcribe Audio\",\n",
+        "    description=(\n",
+        "        \"Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the\"\n",
+        "        \" checkpoint [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) and 🤗 Transformers to transcribe audio files\"\n",
+        "        \" of arbitrary length.\"\n",
+        "    ),\n",
+        "    allow_flagging=\"never\",\n",
+        ")\n",
+        "\n",
+        "demo.launch()\n"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 648
+        },
+        "id": "K0b2UZLVdIze",
+        "outputId": "bcff00e0-4fc8-4883-9ba4-480f5a6665f0"
+      },
+      "execution_count": null,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Setting queue=True in a Colab notebook requires sharing enabled. Setting `share=True` (you can turn this off by setting `share=False` in `launch()` explicitly).\n",
+            "\n",
+            "Colab notebook detected. To show errors in colab notebook, set debug=True in launch()\n",
+            "Running on public URL: https://037dbdb04542aa1a29.gradio.live\n",
+            "\n",
+            "This share link expires in 72 hours. For free permanent hosting and GPU upgrades, run `gradio deploy` from Terminal to deploy to Spaces (https://huggingface.co/spaces)\n"
+          ]
+        },
+        {
+          "output_type": "display_data",
+          "data": {
+            "text/plain": [
+              "<IPython.core.display.HTML object>"
+            ],
+            "text/html": [
+              "<div><iframe src=\"https://037dbdb04542aa1a29.gradio.live\" width=\"100%\" height=\"500\" allow=\"autoplay; camera; microphone; clipboard-read; clipboard-write;\" frameborder=\"0\" allowfullscreen></iframe></div>"
+            ]
+          },
+          "metadata": {}
+        },
+        {
+          "output_type": "execute_result",
+          "data": {
+            "text/plain": []
+          },
+          "metadata": {},
+          "execution_count": 4
+        }
+      ]
+    }
+  ]
+}