mirror of
https://github.com/snakers4/silero-vad.git
synced 2026-02-05 18:09:22 +08:00
collab fx
This commit is contained in:
adamnsandle
202
silero-vad.ipynb
202
silero-vad.ipynb
@@ -1,42 +1,4 @@
|
||||
{
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 0,
|
||||
"metadata": {
|
||||
"colab": {
|
||||
"name": "silero-vad.ipynb",
|
||||
"provenance": []
|
||||
},
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3",
|
||||
"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": "3.8.8"
|
||||
},
|
||||
"toc": {
|
||||
"base_numbering": 1,
|
||||
"nav_menu": {},
|
||||
"number_sections": true,
|
||||
"sideBar": true,
|
||||
"skip_h1_title": false,
|
||||
"title_cell": "Table of Contents",
|
||||
"title_sidebar": "Contents",
|
||||
"toc_cell": false,
|
||||
"toc_position": {},
|
||||
"toc_section_display": true,
|
||||
"toc_window_display": false
|
||||
}
|
||||
},
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -68,15 +30,17 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "5w5AkskZ2Fwr"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#@title Install and Import Dependencies\n",
|
||||
"\n",
|
||||
"# this assumes that you have a relevant version of PyTorch installed\n",
|
||||
"!pip install -q torchaudio soundfile\n",
|
||||
"!pip install -q torchaudio\n",
|
||||
"\n",
|
||||
"SAMPLE_RATE = 16000\n",
|
||||
"\n",
|
||||
@@ -98,9 +62,7 @@
|
||||
" collect_chunks) = utils\n",
|
||||
"\n",
|
||||
"files_dir = torch.hub.get_dir() + '/snakers4_silero-vad_master/files'"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -122,31 +84,31 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"id": "aI_eydBPjsrx"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"wav = read_audio(f'{files_dir}/en.wav', sampling_rate=SAMPLE_RATE)\n",
|
||||
"# get speech timestamps from full audio file\n",
|
||||
"speech_timestamps = get_speech_timestamps(wav, model, sampling_rate=SAMPLE_RATE)\n",
|
||||
"pprint(speech_timestamps)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"id": "OuEobLchjsry"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# merge all speech chunks to one audio\n",
|
||||
"save_audio('only_speech.wav',\n",
|
||||
" collect_chunks(speech_timestamps, wav), sampling_rate=16000) \n",
|
||||
"Audio('only_speech.wav')"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -154,19 +116,21 @@
|
||||
"id": "iDKQbVr8jsry"
|
||||
},
|
||||
"source": [
|
||||
"**Stream imitation example**"
|
||||
"### Stream imitation example"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"id": "q-lql_2Wjsry"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"## using VADIterator class\n",
|
||||
"\n",
|
||||
"vad_iterator = VADiterator(double_model)\n",
|
||||
"wav = read_audio((f'{files_dir}/en.wav', sampling_rate=SAMPLE_RATE)\n",
|
||||
"vad_iterator = VADIterator(model)\n",
|
||||
"wav = read_audio(f'{files_dir}/en.wav', sampling_rate=SAMPLE_RATE)\n",
|
||||
"\n",
|
||||
"window_size_samples = 1536 # number of samples in a single audio chunk\n",
|
||||
"for i in range(0, len(wav), window_size_samples):\n",
|
||||
@@ -174,19 +138,19 @@
|
||||
" if speech_dict:\n",
|
||||
" print(speech_dict, end=' ')\n",
|
||||
"vad_iterator.reset_states() # reset model states after each audio"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"id": "BX3UgwwB2Fwv"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"## just probabilities\n",
|
||||
"\n",
|
||||
"wav = read_audio((f'{files_dir}/en.wav', sampling_rate=SAMPLE_RATE)\n",
|
||||
"wav = read_audio(f'{files_dir}/en.wav', sampling_rate=SAMPLE_RATE)\n",
|
||||
"speech_probs = []\n",
|
||||
"window_size_samples = 1536\n",
|
||||
"for i in range(0, len(wav), window_size_samples):\n",
|
||||
@@ -194,9 +158,7 @@
|
||||
" speech_probs.append(speech_prob)\n",
|
||||
"\n",
|
||||
"pprint(speech_probs[:100])"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -221,10 +183,12 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "Kq5gQuYq2Fwx"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#@title Install and Import Dependencies\n",
|
||||
"\n",
|
||||
@@ -249,9 +213,7 @@
|
||||
" drop_chunks) = utils\n",
|
||||
"\n",
|
||||
"files_dir = torch.hub.get_dir() + '/snakers4_silero-vad_master/files'"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -266,64 +228,64 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "EXpau6xq2Fwy"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"wav = read_audio(f'{files_dir}/en_num.wav')\n",
|
||||
"# get number timestamps from full audio file\n",
|
||||
"number_timestamps = get_number_ts(wav, model)\n",
|
||||
"pprint(number_timestamps)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "u-KfXRhZ2Fwy"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"sample_rate = 16000\n",
|
||||
"# convert ms in timestamps to samples\n",
|
||||
"for timestamp in number_timestamps:\n",
|
||||
" timestamp['start'] = int(timestamp['start'] * sample_rate / 1000)\n",
|
||||
" timestamp['end'] = int(timestamp['end'] * sample_rate / 1000)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "iwYEC4aZ2Fwy"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# merge all number chunks to one audio\n",
|
||||
"save_audio('only_numbers.wav',\n",
|
||||
" collect_chunks(number_timestamps, wav), sample_rate) \n",
|
||||
"Audio('only_numbers.wav')"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "fHaYejX12Fwy"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# drop all number chunks from audio\n",
|
||||
"save_audio('no_numbers.wav',\n",
|
||||
" drop_chunks(number_timestamps, wav), sample_rate) \n",
|
||||
"Audio('no_numbers.wav')"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -348,10 +310,12 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "Zu9D0t6n2Fwz"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#@title Install and Import Dependencies\n",
|
||||
"\n",
|
||||
@@ -373,9 +337,7 @@
|
||||
" read_audio) = utils\n",
|
||||
"\n",
|
||||
"files_dir = torch.hub.get_dir() + '/snakers4_silero-vad_master/files'"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -390,17 +352,17 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "c8UYnYBF2Fw0"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"wav = read_audio(f'{files_dir}/en.wav')\n",
|
||||
"lang = get_language(wav, model)\n",
|
||||
"print(lang)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -452,11 +414,13 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"cellView": "form",
|
||||
"hidden": true,
|
||||
"id": "PdjGd56R2Fw5"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#@title Install and Import Dependencies\n",
|
||||
"\n",
|
||||
@@ -491,9 +455,7 @@
|
||||
" outs = model.run(None, ort_inputs)\n",
|
||||
" outs = [torch.Tensor(x) for x in outs]\n",
|
||||
" return outs"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -508,10 +470,12 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "_r6QZiwu2Fw5"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"model = init_onnx_model(f'{files_dir}/number_detector.onnx')\n",
|
||||
"wav = read_audio(f'{files_dir}/en_num.wav')\n",
|
||||
@@ -519,55 +483,53 @@
|
||||
"# get number timestamps from full audio file\n",
|
||||
"number_timestamps = get_number_ts(wav, model, run_function=validate_onnx)\n",
|
||||
"pprint(number_timestamps)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "FN4aDwLV2Fw5"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"sample_rate = 16000\n",
|
||||
"# convert ms in timestamps to samples\n",
|
||||
"for timestamp in number_timestamps:\n",
|
||||
" timestamp['start'] = int(timestamp['start'] * sample_rate / 1000)\n",
|
||||
" timestamp['end'] = int(timestamp['end'] * sample_rate / 1000)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "JnvS6WTK2Fw5"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# merge all number chunks to one audio\n",
|
||||
"save_audio('only_numbers.wav',\n",
|
||||
" collect_chunks(number_timestamps, wav), 16000) \n",
|
||||
"Audio('only_numbers.wav')"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "yUxOcOFG2Fw6"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# drop all number chunks from audio\n",
|
||||
"save_audio('no_numbers.wav',\n",
|
||||
" drop_chunks(number_timestamps, wav), 16000) \n",
|
||||
"Audio('no_numbers.wav')"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -592,11 +554,13 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"cellView": "form",
|
||||
"hidden": true,
|
||||
"id": "iNkDWJ3H2Fw6"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#@title Install and Import Dependencies\n",
|
||||
"\n",
|
||||
@@ -628,9 +592,7 @@
|
||||
" outs = model.run(None, ort_inputs)\n",
|
||||
" outs = [torch.Tensor(x) for x in outs]\n",
|
||||
" return outs"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
@@ -644,19 +606,57 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"hidden": true,
|
||||
"id": "WHXnh9IV2Fw6"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"model = init_onnx_model(f'{files_dir}/number_detector.onnx')\n",
|
||||
"wav = read_audio(f'{files_dir}/en.wav')\n",
|
||||
"\n",
|
||||
"lang = get_language(wav, model, run_function=validate_onnx)\n",
|
||||
"print(lang)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
}
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"colab": {
|
||||
"name": "silero-vad.ipynb",
|
||||
"provenance": []
|
||||
},
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3",
|
||||
"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": "3.8.8"
|
||||
},
|
||||
"toc": {
|
||||
"base_numbering": 1,
|
||||
"nav_menu": {},
|
||||
"number_sections": true,
|
||||
"sideBar": true,
|
||||
"skip_h1_title": false,
|
||||
"title_cell": "Table of Contents",
|
||||
"title_sidebar": "Contents",
|
||||
"toc_cell": false,
|
||||
"toc_position": {},
|
||||
"toc_section_display": true,
|
||||
"toc_window_display": false
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 0
|
||||
}
|
||||
|
||||
@@ -20,7 +20,6 @@ def validate(model,
|
||||
def read_audio(path: str,
|
||||
sampling_rate: int = 16000):
|
||||
|
||||
assert torchaudio.get_audio_backend() == 'soundfile'
|
||||
wav, sr = torchaudio.load(path)
|
||||
|
||||
if wav.size(0) > 1:
|
||||
|
||||
Reference in New Issue
Block a user