232 lines
7.9 KiB
Plaintext
232 lines
7.9 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "eba9e610",
|
|
"metadata": {},
|
|
"source": [
|
|
"A simple way to avoid being connected while transcribing is to first load the model version you want to use. See [here](https://github.com/openai/whisper/blob/main/README.md#available-models-and-languages) for more info."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 6,
|
|
"id": "85cd2d12",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"Whisper(\n",
|
|
" (encoder): AudioEncoder(\n",
|
|
" (conv1): Conv1d(80, 1024, kernel_size=(3,), stride=(1,), padding=(1,))\n",
|
|
" (conv2): Conv1d(1024, 1024, kernel_size=(3,), stride=(2,), padding=(1,))\n",
|
|
" (blocks): ModuleList(\n",
|
|
" (0-23): 24 x ResidualAttentionBlock(\n",
|
|
" (attn): MultiHeadAttention(\n",
|
|
" (query): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" (key): Linear(in_features=1024, out_features=1024, bias=False)\n",
|
|
" (value): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" (out): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" )\n",
|
|
" (attn_ln): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
|
|
" (mlp): Sequential(\n",
|
|
" (0): Linear(in_features=1024, out_features=4096, bias=True)\n",
|
|
" (1): GELU(approximate='none')\n",
|
|
" (2): Linear(in_features=4096, out_features=1024, bias=True)\n",
|
|
" )\n",
|
|
" (mlp_ln): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
|
|
" )\n",
|
|
" )\n",
|
|
" (ln_post): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
|
|
" )\n",
|
|
" (decoder): TextDecoder(\n",
|
|
" (token_embedding): Embedding(51865, 1024)\n",
|
|
" (blocks): ModuleList(\n",
|
|
" (0-23): 24 x ResidualAttentionBlock(\n",
|
|
" (attn): MultiHeadAttention(\n",
|
|
" (query): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" (key): Linear(in_features=1024, out_features=1024, bias=False)\n",
|
|
" (value): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" (out): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" )\n",
|
|
" (attn_ln): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
|
|
" (cross_attn): MultiHeadAttention(\n",
|
|
" (query): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" (key): Linear(in_features=1024, out_features=1024, bias=False)\n",
|
|
" (value): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" (out): Linear(in_features=1024, out_features=1024, bias=True)\n",
|
|
" )\n",
|
|
" (cross_attn_ln): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
|
|
" (mlp): Sequential(\n",
|
|
" (0): Linear(in_features=1024, out_features=4096, bias=True)\n",
|
|
" (1): GELU(approximate='none')\n",
|
|
" (2): Linear(in_features=4096, out_features=1024, bias=True)\n",
|
|
" )\n",
|
|
" (mlp_ln): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
|
|
" )\n",
|
|
" )\n",
|
|
" (ln): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
|
|
" )\n",
|
|
")"
|
|
]
|
|
},
|
|
"execution_count": 6,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"import whisper\n",
|
|
"#change to model size, bigger is more accurate but slower\n",
|
|
"whisper.load_model(\"medium\") #base, small, medium, large"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 7,
|
|
"id": "0d2acd54",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"#after it loads, you can disconnect from the internet and run the rest"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 8,
|
|
"id": "a2cd4050",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"from transcribe import transcribe"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 9,
|
|
"id": "24e1d24e",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"Help on function transcribe in module transcribe:\n",
|
|
"\n",
|
|
"transcribe(path, file_type, model=None, language=None, verbose=True)\n",
|
|
" Implementation of OpenAI's whisper model. Downloads model, transcribes audio files in a folder and returns the text files with transcriptions\n",
|
|
"\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"help(transcribe)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 11,
|
|
"id": "e52477fb",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"path='sample_audio/'#folder path\n",
|
|
"file_type='ogg' #check your file for file type, will only transcribe files with the file type, 'ogg', 'WAV'\n",
|
|
"model='medium' #'small', 'medium', 'large' (tradeoff between speed and accuracy)\n",
|
|
"language= None #tries to auto-detect, other options include 'English', 'Spanish', etc...\n",
|
|
"verbose = True # prints output while transcribing, False to deactivate"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 12,
|
|
"id": "d66866af",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"Using medium model, you can change this by specifying model=\"medium\" for example\n",
|
|
"Only looking for file type ogg, you can change this by specifying file_type=\"mp3\"\n",
|
|
"Expecting None language, you can change this by specifying language=\"English\". None will try to auto-detect\n",
|
|
"Verbosity is True. If TRUE it will print out the text as it is transcribed, you can turn this off by setting verbose=False\n",
|
|
"\n",
|
|
"There are 2 ogg files in path: sample_audio/\n",
|
|
"\n",
|
|
"\n",
|
|
"Loading model...\n",
|
|
"Transcribing file number number 1: Armstrong_Small_Step\n",
|
|
"Model and file loaded...\n",
|
|
"Starting transcription...\n",
|
|
"\n",
|
|
"Detecting language using up to the first 30 seconds. Use `--language` to specify the language\n",
|
|
"Detected language: English\n",
|
|
"[00:00.000 --> 00:24.000] That's one small step for man, one giant leap for mankind.\n",
|
|
"\n",
|
|
"Finished file number 1.\n",
|
|
"\n",
|
|
"\n",
|
|
"\n",
|
|
"Transcribing file number number 2: Axel_Pettersson_röstinspelning\n",
|
|
"Model and file loaded...\n",
|
|
"Starting transcription...\n",
|
|
"\n",
|
|
"Detecting language using up to the first 30 seconds. Use `--language` to specify the language\n",
|
|
"Detected language: Swedish\n",
|
|
"[00:00.000 --> 00:16.000] Hej, jag heter Axel Pettersson, jag föddes i Örebro 1976. Jag har varit Wikipedia sen 2008 och jag har översatt röstintroduktionsprojektet till svenska.\n",
|
|
"\n",
|
|
"Finished file number 2.\n",
|
|
"\n",
|
|
"\n",
|
|
"\n"
|
|
]
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"'Finished transcription, files can be found in sample_audio/transcriptions'"
|
|
]
|
|
},
|
|
"execution_count": 12,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"transcribe(path, file_type, model, language, verbose)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"id": "0bc67265",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": []
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": {
|
|
"display_name": "Python 3 (ipykernel)",
|
|
"language": "python",
|
|
"name": "python3"
|
|
},
|
|
"language_info": {
|
|
"codemirror_mode": {
|
|
"name": "ipython",
|
|
"version": 3
|
|
},
|
|
"file_extension": ".py",
|
|
"mimetype": "text/x-python",
|
|
"name": "python",
|
|
"nbconvert_exporter": "python",
|
|
"pygments_lexer": "ipython3",
|
|
"version": "3.10.4"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 5
|
|
}
|