mirror of https://github.com/coqui-ai/TTS.git
Clean-up `CheckSpectrograms` Notebook
- Fixes how parameters from the config are loaded while still making it possible for them to be overridden on the fly. - Prints the list of audio files by index. - Defines a constant to control which audio files gets loaded for all sections. - Eliminates having to load the audio processor twice. - Removes output (since it isn't relevant).pull/393/head
Guy Elsmore-Paddock
parent
0601f6fc0f
commit
49484f0937
|
|
@ -2,11 +2,9 @@
|
|||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"%matplotlib inline\n",
|
||||
"\n",
|
||||
|
|
@ -16,21 +14,32 @@
|
|||
"\n",
|
||||
"import IPython.display as ipd\n",
|
||||
"import glob"
|
||||
]
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"config_path = \"/home/erogol/Projects/TTS/tts/tts/config_thorsten_de.json\"\n",
|
||||
"data_path = \"/home/erogol/Data/thorsten-german/\"\n",
|
||||
"\n",
|
||||
"file_paths = glob.glob(data_path + \"/**/*.wav\", recursive=True)\n",
|
||||
"CONFIG = load_config(config_path)"
|
||||
]
|
||||
"CONFIG = load_config(config_path)\n",
|
||||
"\n",
|
||||
"# Change this to the index of the desired file listed below\n",
|
||||
"sample_file_index = 10\n",
|
||||
"\n",
|
||||
"SAMPLE_FILE_PATH = file_paths[sample_file_index]\n",
|
||||
"\n",
|
||||
"print(\"File list, by index:\")\n",
|
||||
"dict(enumerate(file_paths))"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
|
|
@ -39,39 +48,54 @@
|
|||
},
|
||||
"source": [
|
||||
"### Setup Audio Processor\n",
|
||||
"Play with the AP parameters until you find a good fit with the synthesis speech below. "
|
||||
"Play with the AP parameters until you find a good fit with the synthesis speech below.\n",
|
||||
"\n",
|
||||
"The default values are loaded from your config.json file, so you only need to\n",
|
||||
"uncomment and modify values below that you'd like to tune."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# audio={\n",
|
||||
"tune_params={\n",
|
||||
"# 'audio_processor': 'audio',\n",
|
||||
"# 'num_mels': 80, # In general, you don'tneed to change it \n",
|
||||
"# 'fft_size': 1024, # In general, you don'tneed to change it \n",
|
||||
"# 'sample_rate': 22050, # It depends to the sample rate of the dataset.\n",
|
||||
"# 'hop_length': 256, # In general, you don'tneed to change it \n",
|
||||
"# 'win_length': 1024, # In general, you don'tneed to change it \n",
|
||||
"# 'preemphasis': 0.98, # In general, 0 gives better voice recovery but makes traning harder. If your model does not train, try 0.97 - 0.99.\n",
|
||||
"# 'num_mels': 80, # In general, you don't need to change this. \n",
|
||||
"# 'fft_size': 1024, # In general, you don't need to change this.\n",
|
||||
"# 'sample_rate': 22050, # This must match the sample rate of the dataset.\n",
|
||||
"# 'hop_length': 256, # In general, you don't need to change this.\n",
|
||||
"# 'win_length': 1024, # In general, you don't need to change this.\n",
|
||||
"# 'preemphasis': 0.98, # In general, 0 gives better voice recovery but makes training harder. If your model does not train, try 0.97 - 0.99.\n",
|
||||
"# 'min_level_db': -100,\n",
|
||||
"# 'ref_level_db': 20, # It is the base DB, higher until you remove the background noise in the spectrogram and then lower until you hear a better speech below.\n",
|
||||
"# 'power': 1.5, # Change this value and listen the synthesized voice. 1.2 - 1.5 are some resonable values.\n",
|
||||
"# 'griffin_lim_iters': 60, # It does not give any imporvement for values > 60\n",
|
||||
"# 'signal_norm': True, # This is more about your model. It does not give any change for the synthsis performance.\n",
|
||||
"# 'symmetric_norm': False, # Same as above\n",
|
||||
"# 'max_norm': 1, # Same as above\n",
|
||||
"# 'clip_norm': True, # Same as above\n",
|
||||
"# 'mel_fmin': 0.0, # You can play with this and check mel-spectrogram based voice synthesis below.\n",
|
||||
"# 'mel_fmax': 8000.0, # You can play with this and check mel-spectrogram based voice synthesis below.\n",
|
||||
"# 'do_trim_silence': True} # If you dataset has some silience at the beginning or end, this trims it. Check the AP.load_wav() below,if it causes any difference for the loaded audio file.\n",
|
||||
"# 'ref_level_db': 0, # The base DB; increase until all background noise is removed in the spectrogram, then lower until you hear better speech below.\n",
|
||||
"# 'power': 1.5, # Change this value and listen to the synthesized voice. 1.2 - 1.5 are resonable values.\n",
|
||||
"# 'griffin_lim_iters': 60, # Quality does not improve for values > 60\n",
|
||||
"# 'mel_fmin': 0.0, # Adjust this and check mel-spectrogram-based voice synthesis below.\n",
|
||||
"# 'mel_fmax': 8000.0, # Adjust this and check mel-spectrogram-based voice synthesis below.\n",
|
||||
"# 'do_trim_silence': True # If you dataset has some silience at the beginning or end, this trims it. Check the AP.load_wav() below,if it causes any difference for the loaded audio file.\n",
|
||||
"}\n",
|
||||
"\n",
|
||||
"AP = AudioProcessor(**CONFIG.audio);"
|
||||
]
|
||||
"# These options have to be forced off in order to avoid errors about the \n",
|
||||
"# pre-calculated not matching the options being tuned.\n",
|
||||
"reset={\n",
|
||||
" 'signal_norm': False,\n",
|
||||
" 'stats_path': None,\n",
|
||||
" 'symmetric_norm': False,\n",
|
||||
" 'max_norm': 1,\n",
|
||||
" 'clip_norm': True,\n",
|
||||
"}\n",
|
||||
"\n",
|
||||
"# Override select parts of loaded config with parameters above\n",
|
||||
"tuned_config = CONFIG.audio.copy()\n",
|
||||
"tuned_config.update(reset)\n",
|
||||
"tuned_config.update(tune_params)\n",
|
||||
"\n",
|
||||
"AP = AudioProcessor(**tuned_config);"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
|
|
@ -84,15 +108,15 @@
|
|||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"wav = AP.load_wav(file_paths[10])\n",
|
||||
"wav = AP.load_wav(SAMPLE_FILE_PATH)\n",
|
||||
"ipd.Audio(data=wav, rate=AP.sample_rate) "
|
||||
]
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
|
|
@ -105,20 +129,17 @@
|
|||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"AP.power = 1.0"
|
||||
]
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"mel = AP.melspectrogram(wav)\n",
|
||||
"print(\"Max:\", mel.max())\n",
|
||||
|
|
@ -128,7 +149,9 @@
|
|||
"\n",
|
||||
"wav_gen = AP.inv_melspectrogram(mel)\n",
|
||||
"ipd.Audio(wav_gen, rate=AP.sample_rate)"
|
||||
]
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
|
|
@ -141,11 +164,9 @@
|
|||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"spec = AP.spectrogram(wav)\n",
|
||||
"print(\"Max:\", spec.max())\n",
|
||||
|
|
@ -155,7 +176,9 @@
|
|||
"\n",
|
||||
"wav_gen = AP.inv_spectrogram(spec)\n",
|
||||
"ipd.Audio(wav_gen, rate=AP.sample_rate)"
|
||||
]
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
|
|
@ -170,54 +193,21 @@
|
|||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"audio={\n",
|
||||
" 'audio_processor': 'audio',\n",
|
||||
" 'num_mels': 80, # In general, you don'tneed to change it \n",
|
||||
" 'num_freq': 1025, # In general, you don'tneed to change it \n",
|
||||
" 'sample_rate': 22050, # It depends to the sample rate of the dataset.\n",
|
||||
" 'frame_length_ms': 50, # In general, you don'tneed to change it \n",
|
||||
" 'frame_shift_ms': 12.5, # In general, you don'tneed to change it \n",
|
||||
" 'preemphasis': 0.98, # In general, 0 gives better voice recovery but makes traning harder. If your model does not train, try 0.97 - 0.99.\n",
|
||||
" 'min_level_db': -100,\n",
|
||||
" 'ref_level_db': 20, # It is the base DB, higher until you remove the background noise in the spectrogram and then lower until you hear a better speech below.\n",
|
||||
" 'power': 1.5, # Change this value and listen the synthesized voice. 1.2 - 1.5 are some resonable values.\n",
|
||||
" 'griffin_lim_iters': 60, # It does not give any imporvement for values > 60\n",
|
||||
" 'signal_norm': True, # This is more about your model. It does not give any change for the synthsis performance.\n",
|
||||
" 'symmetric_norm': False, # Same as above\n",
|
||||
" 'max_norm': 1, # Same as above\n",
|
||||
" 'clip_norm': True, # Same as above\n",
|
||||
" 'mel_fmin': 0.0, # You can play with this and check mel-spectrogram based voice synthesis below.\n",
|
||||
" 'mel_fmax': 8000.0, # You can play with this and check mel-spectrogram based voice synthesis below.\n",
|
||||
" 'do_trim_silence': True} # If you dataset has some silience at the beginning or end, this trims it. Check the AP.load_wav() below,if it causes any difference for the loaded audio file.\n",
|
||||
"\n",
|
||||
"AP = AudioProcessor(**audio);"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from librosa import display\n",
|
||||
"from matplotlib import pylab as plt\n",
|
||||
"import IPython\n",
|
||||
"plt.rcParams['figure.figsize'] = (20.0, 16.0)\n",
|
||||
"\n",
|
||||
"def compare_values(attribute, values, file):\n",
|
||||
"def compare_values(attribute, values):\n",
|
||||
" \"\"\"\n",
|
||||
" attributes (str): the names of the attribute you like to test.\n",
|
||||
" values (list): list of values to compare.\n",
|
||||
" file (str): file name to perform the tests.\n",
|
||||
" \"\"\"\n",
|
||||
" file = SAMPLE_FILE_PATH\n",
|
||||
" wavs = []\n",
|
||||
" for idx, val in enumerate(values):\n",
|
||||
" set_val_cmd = \"AP.{}={}\".format(attribute, val)\n",
|
||||
|
|
@ -245,29 +235,31 @@
|
|||
" val = values[idx]\n",
|
||||
" print(\" > {} = {}\".format(attribute, val))\n",
|
||||
" IPython.display.display(IPython.display.Audio(wav_gen, rate=AP.sample_rate))"
|
||||
]
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"compare_values(\"preemphasis\", [0, 0.5, 0.97, 0.98, 0.99], file_paths[10])"
|
||||
]
|
||||
"compare_values(\"preemphasis\", [0, 0.5, 0.97, 0.98, 0.99])"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"Collapsed": "false"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"compare_values(\"ref_level_db\", [10, 15, 20, 25, 30, 35, 40], file_paths[10])"
|
||||
]
|
||||
"compare_values(\"ref_level_db\", [2, 5, 10, 15, 20, 25, 30, 35, 40, 100])"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
|
|
|
|||
Loading…
Reference in New Issue