Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

🦸 Update notebook and gradio code #111

Merged
merged 1 commit into from
Jun 8, 2024
Merged

🦸 Update notebook and gradio code #111

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
177 changes: 154 additions & 23 deletions notebook/demo.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,17 @@
"nest_asyncio.apply()"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "9b6cf901",
"metadata": {},
"outputs": [],
"source": [
"!pip install whisperplus git+https://github.com/huggingface/transformers\n",
"!pip install flash-attn --no-build-isolation"
]
},
{
"cell_type": "markdown",
"id": "1d0b2e40",
Expand All @@ -50,16 +61,100 @@
"metadata": {},
"outputs": [],
"source": [
"from whisperplus import SpeechToTextPipeline, download_and_convert_to_mp3\n",
"from whisperplus import SpeechToTextPipeline, download_youtube_to_mp3\n",
"from transformers import BitsAndBytesConfig, HqqConfig\n",
"import torch\n",
"\n",
"url = \"https://www.youtube.com/watch?v=di3rHkEZuUw\"\n",
"audio_path = download_and_convert_to_mp3(url)\n",
"pipeline = SpeechToTextPipeline(model_id=\"openai/whisper-large-v3\")\n",
"transcript = pipeline(audio_path, \"openai/whisper-large-v3\", \"english\")\n",
"audio_path = download_youtube_to_mp3(url, output_dir=\"downloads\", filename=\"test\")\n",
"\n",
"hqq_config = HqqConfig(\n",
" nbits=4,\n",
" group_size=64,\n",
" quant_zero=False,\n",
" quant_scale=False,\n",
" axis=0,\n",
" offload_meta=False,\n",
") # axis=0 is used by default\n",
"\n",
"bnb_config = BitsAndBytesConfig(\n",
" load_in_4bit=True,\n",
" bnb_4bit_quant_type=\"nf4\",\n",
" bnb_4bit_compute_dtype=torch.bfloat16,\n",
" bnb_4bit_use_double_quant=True,\n",
")\n",
"\n",
"pipeline = SpeechToTextPipeline(\n",
" model_id=\"distil-whisper/distil-large-v3\",\n",
" quant_config=hqq_config,\n",
" flash_attention_2=True,\n",
")\n",
"\n",
"transcript = pipeline(\n",
" audio_path=audio_path,\n",
" chunk_length_s=30,\n",
" stride_length_s=5,\n",
" max_new_tokens=128,\n",
" batch_size=100,\n",
" language=\"english\",\n",
" return_timestamps=False,\n",
")\n",
"\n",
"print(transcript)"
]
},
{
"cell_type": "markdown",
"id": "8d6282f7",
"metadata": {},
"source": [
"### 🍎 Apple MLX"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "97b42087",
"metadata": {},
"outputs": [],
"source": [
"from whisperplus.pipelines import mlx_whisper\n",
"from whisperplus import download_youtube_to_mp3\n",
"\n",
"url = \"https://www.youtube.com/watch?v=1__CAdTJ5JU\"\n",
"audio_path = download_youtube_to_mp3(url)\n",
"\n",
"text = mlx_whisper.transcribe(\n",
" audio_path, path_or_hf_repo=\"mlx-community/whisper-large-v3-mlx\"\n",
")[\"text\"]\n",
"print(text)"
]
},
{
"cell_type": "markdown",
"id": "ca528ba7",
"metadata": {},
"source": [
"### 🍏 Lightning Mlx Whisper"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "0d99b8bb",
"metadata": {},
"outputs": [],
"source": [
"from whisperplus.pipelines.lightning_whisper_mlx import LightningWhisperMLX\n",
"from whisperplus import download_youtube_to_mp3\n",
"\n",
"url = \"https://www.youtube.com/watch?v=1__CAdTJ5JU\"\n",
"audio_path = download_youtube_to_mp3(url)\n",
"\n",
"whisper = LightningWhisperMLX(model=\"distil-large-v3\", batch_size=12, quant=None)\n",
"output = whisper.transcribe(audio_path=audio_path)[\"text\"]"
]
},
{
"cell_type": "markdown",
"id": "1a99131f",
Expand All @@ -77,7 +172,7 @@
"metadata": {},
"outputs": [],
"source": [
"from whisperplus import TextSummarizationPipeline\n",
"from whisperplus.pipelines.summarization import TextSummarizationPipeline\n",
"\n",
"summarizer = TextSummarizationPipeline(model_id=\"facebook/bart-large-cnn\")\n",
"summary = summarizer.summarize(transcript)\n",
Expand All @@ -101,7 +196,7 @@
"metadata": {},
"outputs": [],
"source": [
"from whisperplus import LongTextSummarizationPipeline\n",
"from whisperplus.pipelines.long_text_summarization import LongTextSummarizationPipeline\n",
"\n",
"summarizer = LongTextSummarizationPipeline(model_id=\"facebook/bart-large-cnn\")\n",
"summary_text = summarizer.summarize(transcript)\n",
Expand All @@ -115,7 +210,25 @@
"source": [
"### 💬 Speaker Diarization\n",
"\n",
"In this section, we demonstrate the use of Speaker Diarization. This feature helps in distinguishing between different speakers in an audio clip."
"You must confirm the licensing permissions of these two models.\n",
"\n",
"- https://huggingface.co/pyannote/speaker-diarization-3.1\n",
"- https://huggingface.co/pyannote/segmentation-3.0"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "0066de25",
"metadata": {},
"outputs": [],
"source": [
"huggingface_hub import notebook_login\n",
"\n",
"!pip install -r speaker_diarization.txt\n",
"!pip install -U \"huggingface_hub[cli]\"\n",
"\n",
"notebook_login()"
]
},
{
Expand All @@ -125,18 +238,15 @@
"metadata": {},
"outputs": [],
"source": [
"from whisperplus import (\n",
" ASRDiarizationPipeline,\n",
" download_and_convert_to_mp3,\n",
" format_speech_to_dialogue,\n",
")\n",
"from whisperplus.pipelines.whisper_diarize import ASRDiarizationPipeline\n",
"from whisperplus import download_youtube_to_mp3, format_speech_to_dialogue\n",
"\n",
"audio_path = download_and_convert_to_mp3(\"https://www.youtube.com/watch?v=mRB14sFHw2E\")\n",
"audio_path = download_youtube_to_mp3(\"https://www.youtube.com/watch?v=mRB14sFHw2E\")\n",
"\n",
"device = \"cuda\" # cpu or mps\n",
"pipeline = ASRDiarizationPipeline.from_pretrained(\n",
" asr_model=\"openai/whisper-large-v3\",\n",
" diarizer_model=\"pyannote/speaker-diarization\",\n",
" diarizer_model=\"pyannote/speaker-diarization-3.1\",\n",
" use_auth_token=False,\n",
" chunk_length_s=30,\n",
" device=device,\n",
Expand All @@ -157,14 +267,24 @@
"This part covers the 'Chat with Video' feature using LanceDB. It demonstrates how to interact with a video transcript using a chat interface."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "0714c378",
"metadata": {},
"outputs": [],
"source": [
"!pip install sentence-transformers ctransformers langchain"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "7c108aee",
"metadata": {},
"outputs": [],
"source": [
"from whisperplus import ChatWithVideo\n",
"from whisperplus.pipelines.chatbot import ChatWithVideo\n",
"\n",
"chat = ChatWithVideo(\n",
" input_file=\"trascript.txt\",\n",
Expand All @@ -175,7 +295,8 @@
")\n",
"\n",
"query = \"what is this video about ?\"\n",
"response = chat.run_query(query)"
"response = chat.run_query(query)\n",
"print(response)"
]
},
{
Expand All @@ -188,14 +309,24 @@
"This section demonstrates the 'Chat with Video' feature using AutoLLM. It enables querying a video's content through a chat interface, utilizing advanced language models."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "38be611e",
"metadata": {},
"outputs": [],
"source": [
"!pip install autollm>=0.1.9"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "ddfd23fd",
"metadata": {},
"outputs": [],
"source": [
"from whisperplus import AutoLLMChatWithVideo\n",
"from whisperplus.pipelines.autollm_chatbot import AutoLLMChatWithVideo\n",
"\n",
"# service_context_params\n",
"system_prompt = \"\"\"\n",
Expand All @@ -215,15 +346,15 @@
"\"\"\"\n",
"\n",
"chat = AutoLLMChatWithVideo(\n",
" input_file=\"audio.mp3\",\n",
" openai_key=\"YOUR_OPENAI_KEY\",\n",
" huggingface_key=\"YOUR_HUGGINGFACE_KEY\",\n",
" input_file=\"input_dir\", # path of mp3 file\n",
" openai_key=\"YOUR_OPENAI_KEY\", # optional\n",
" huggingface_key=\"YOUR_HUGGINGFACE_KEY\", # optional\n",
" llm_model=\"gpt-3.5-turbo\",\n",
" llm_max_tokens=\"256\",\n",
" llm_temperature=\"0.1\",\n",
" system_prompt=system_prompt,\n",
" query_wrapper_prompt=query_wrapper_prompt,\n",
" embed_model=\"huggingface/BAAI/bge-large-zh\",\n",
" embed_model=\"huggingface/BAAI/bge-large-zh\", # \"text-embedding-ada-002\"\n",
")\n",
"\n",
"query = \"what is this video about ?\"\n",
Expand All @@ -236,7 +367,7 @@
"id": "223ed48e",
"metadata": {},
"source": [
"### 🎙️ Speech to Text\n",
"### 🎙️ Text to Speech\n",
"\n",
"Finally, this section covers converting text to speech using WhisperPlus, demonstrating how to generate spoken audio from text."
]
Expand All @@ -248,7 +379,7 @@
"metadata": {},
"outputs": [],
"source": [
"from whisperplus import TextToSpeechPipeline\n",
"from whisperplus.pipelines.text2speech import TextToSpeechPipeline\n",
"\n",
"tts = TextToSpeechPipeline(model_id=\"suno/bark\")\n",
"audio = tts(text=\"Hello World\", voice_preset=\"v2/en_speaker_6\")"
Expand Down
Loading
Loading