File size: 5,769 Bytes
d9edac2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 | {
"cells": [
{
"cell_type": "markdown",
"source": [
"# HebTTS Quickstart\n",
"We created a jupyter notebook in order to easily generate your desired samples!\n",
"\n",
"\n",
" <a href='https://arxiv.org/abs/2407.12206'><img src='https://img.shields.io/badge/ArXiv-PDF-red'></a> \n",
" <a href='https://pages.cs.huji.ac.il/adiyoss-lab/HebTTS/'><img src='https://img.shields.io/badge/Project-Page-Green'></a> \n",
" [](https://colab.research.google.com/drive/1f3-6Dqbna9_hI5C9V4qTIG05dixW-r72?usp=sharing) \n",
" [](https://github.com/slp-rl/HebTTS) \n",
"\n",
"\n",
"\n",
"\n",
"---\n",
"\n",
"\n"
],
"metadata": {
"id": "3wmzNy2fDSsr"
}
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "56kkbeJahIe1"
},
"outputs": [],
"source": [
"#@title Install โ๐ป\n",
"\n",
"%%capture\n",
"! git clone https://github.com/slp-rl/HebTTS.git\n",
"\n",
"! pip install torch torchaudio\n",
"! pip install torchmetrics\n",
"! pip install omegaconf\n",
"! pip install git+https://github.com/lhotse-speech/lhotse\n",
"! pip install librosa\n",
"! pip install encodec\n",
"! pip install phonemizer\n",
"! pip install audiocraft # optional\n",
"! gdown https://drive.google.com/uc?id=11NoOJzMLRX9q1C_Q4sX0w2b9miiDjGrv\n",
"\n",
"from pathlib import Path\n",
"import glob\n",
"import os\n",
"from IPython.display import Audio, display\n",
"from pathlib import Path\n",
"\n",
"\n",
"def display_audio(prompt_file):\n",
" l=100\n",
" speaker = os.path.basename(os.path.dirname((os.path.dirname(prompt_file))))\n",
" audio_files = list(Path(os.path.join(os.path.dirname(prompt_file))).rglob(\"*.wav\"))\n",
" print(f\"Prompt: {open(prompt_file).read()}\")\n",
" print(f\"Speaker: {speaker}\")\n",
" print(\"-\"*100)\n",
"\n",
" for audio_file in audio_files:\n",
" display(Audio(audio_file, autoplay=False))\n",
" print(\"-\"*100)\n",
" print()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "uIBF-IFwxLjL",
"cellView": "form"
},
"outputs": [],
"source": [
"# @title Enter Hebrew text for generation\n",
"# @markdown You can concatenate prompts with '|' to generate few samples at once\n",
"hebrew_text_to_generate = \"ืืื ืื ืงืืจื | ืืืฉืืื ืืืืื ืืื ืืืืจ ืืืืง ืื ืื ืขืื ืฆืจืื ืืืืืจ ืฉื ืชืืื ืืืจืื ืืช\" # @param {type:\"string\"}"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "RFrYpBJsxQW3",
"cellView": "form"
},
"outputs": [],
"source": [
"\n",
"# @title Choose speaker\n",
"speaker = \"shaul\" # @param [\"\\\"osim\\\"\", \"\\\"geek\\\"\", \"\\\"shaul\\\"\"] {type:\"raw\", allow-input: true}"
]
},
{
"cell_type": "markdown",
"source": [
"**Note:** inference may be slower based on the allocated resources by google colab. It is recomended to choose a gpu in `Edit` -> `Notebook settings` -> `Hardware accelerator`."
],
"metadata": {
"id": "NMvd3VytEJEZ"
}
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "U8Hszx6axgns"
},
"outputs": [],
"source": [
"#@title Generate!\n",
"from pathlib import Path\n",
"\n",
"speaker_dir = Path(f\"./out/{speaker}\")\n",
"speaker_dir.mkdir(exist_ok=True, parents=True)\n",
"i=0\n",
"if any(speaker_dir.iterdir()):\n",
" max(map(lambda x:int(x.name),speaker_dir.glob(\"*\")))+1\n",
"output_dir = speaker_dir / f\"{i}\"\n",
"output_dir.mkdir(exist_ok=True, parents=True)\n",
"prompt_file = output_dir / \"prompt.txt\"\n",
"with open(prompt_file, \"w\") as f:\n",
" f.write(hebrew_text_to_generate)\n",
"\n",
"! python HebTTS/infer.py \\\n",
" --checkpoint checkpoint.pt \\\n",
" --output-dir $output_dir \\\n",
" --text \"$hebrew_text_to_generate\" \\\n",
" --speaker \"$speaker\" \\\n",
" --speaker-yaml HebTTS/speakers/speakers.yaml \\\n",
" --tokens-file HebTTS/tokenizer/unique_words_tokens_all.k2symbols \\\n",
" --vocab-file HebTTS/tokenizer/vocab.txt \\\n",
" --mbd True\n",
"\n",
"display_audio(prompt_file)"
]
},
{
"cell_type": "code",
"source": [
"#@title Listen\n",
"for prompt_file in glob.glob(\"./out/*/*/prompt.txt\"):\n",
" display_audio(prompt_file)"
],
"metadata": {
"cellView": "form",
"id": "CXiMWmNxJzNZ"
},
"execution_count": null,
"outputs": []
}
],
"metadata": {
"colab": {
"provenance": [],
"gpuType": "T4"
},
"kernelspec": {
"display_name": "Python 3",
"name": "python3"
},
"language_info": {
"name": "python"
},
"accelerator": "GPU"
},
"nbformat": 4,
"nbformat_minor": 0
} |