{ "cells": [ { "cell_type": "code", "execution_count": 10, "id": "b2e89c9b", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Populating the interactive namespace from numpy and matplotlib\n" ] } ], "source": [ "%pylab inline" ] }, { "cell_type": "code", "execution_count": 50, "id": "2f8b1a5e", "metadata": {}, "outputs": [], "source": [ "import re\n", "import pandas as pd\n", "import numpy as np\n", "import librosa\n", "from scipy.io import wavfile\n", "import librosa.display\n", "from pydub import AudioSegment\n", "from pydub.utils import mediainfo\n", "from IPython.display import display_html\n", "import torch\n", "import json\n", "import tqdm\n", "\n", "\n", "def _array_to_numpy(arr):\n", " # convert to np if torch\n", " if isinstance(arr, torch.Tensor):\n", " arr = arr.detach().cpu().numpy()\n", " # squash extra dimensions\n", " arr = arr.squeeze()\n", " # TODO: deal with 2 channels\n", " if 2 in arr.shape:\n", " raise NotImplementedError(\"stereo detected\")\n", " return arr\n", " \n", " \n", "def _array_to_float(arr):\n", " arr = _array_to_numpy(arr)\n", " arr = arr = arr.copy().astype(np.float32) / 32768\n", " return arr\n", "\n", "\n", "def _array_to_int(arr):\n", " arr = _array_to_numpy(arr)\n", " # enforce int16\n", " if arr.dtype == np.int16:\n", " pass\n", " elif arr.dtype in (np.float32, np.float64):\n", " # alert if signal too high\n", " if np.abs(arr).max() > 1:\n", " raise ValueError(\"signal overflow\")\n", " arr = (arr * 32768).clip(-32768, 32767).astype(np.int16)\n", " else:\n", " # TODO: convert other formats\n", " raise NotImplementedError(\"unknown format\")\n", " # alert if audio too loud\n", " if np.abs(arr).mean() / 32768 >= 0.1:\n", " raise ValueError(\"audio seems too loud\")\n", " return arr\n", "\n", "\n", "def play_array(arr, sr=16000): \n", " arr = _array_to_int(arr)\n", " display(AudioSegment(\n", " arr.tobytes(), \n", " frame_rate=sr,\n", " sample_width=2, \n", " channels=1\n", " ))" ] }, { "cell_type": "markdown", "id": "ada35f4a", "metadata": {}, "source": [ "## LJSpeech" ] }, { "cell_type": "code", "execution_count": 3, "id": "94b38947", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "alignments\t\t\t mappings.json\r\n", "cmudict.dict\t\t\t metadata.csv\r\n", "energies\t\t\t phoneme_durations\r\n", "ljs_audio_text_test_filelist.txt pitches\r\n", "ljs_audio_text_train_filelist.txt README\r\n", "ljs_audio_text_val_filelist.txt uncommented_cmudict.dict\r\n", "ljspeech_test.json\t\t wavs\r\n", "ljspeech_train.json\t\t wavs_to_ignore.pkl\r\n", "ljspeech_val.json\r\n" ] } ], "source": [ "!ls /mnt/data-ssd-1/data/ljspeech/LJSpeech-1.1" ] }, { "cell_type": "code", "execution_count": 14, "id": "15c13c21", "metadata": {}, "outputs": [], "source": [ "DATA_DIR = \"/mnt/data-ssd-1/data/ljspeech/LJSpeech-1.1/\"" ] }, { "cell_type": "code", "execution_count": 26, "id": "ed828cac", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "22.8 hours total\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
audio_filepathdurationtext
0/mnt/data-ssd-1/data/ljspeech/LJSpeech-1.1/wav...9.863991it has used other treasury law enforcement age...
1/mnt/data-ssd-1/data/ljspeech/LJSpeech-1.1/wav...10.003311to avail himself of his powers, as it was diff...
\n", "
" ], "text/plain": [ " audio_filepath duration \\\n", "0 /mnt/data-ssd-1/data/ljspeech/LJSpeech-1.1/wav... 9.863991 \n", "1 /mnt/data-ssd-1/data/ljspeech/LJSpeech-1.1/wav... 10.003311 \n", "\n", " text \n", "0 it has used other treasury law enforcement age... \n", "1 to avail himself of his powers, as it was diff... " ] }, "execution_count": 26, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df = pd.read_json(DATA_DIR + \"ljspeech_train.json\", lines=True)\n", "print(round(df['duration'].sum() / 60 / 60, 1), \"hours total\")\n", "df.head(2)" ] }, { "cell_type": "code", "execution_count": 27, "id": "77160014", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "all the tumultuous and diversified passions and emotions which circumstances like these must necessarily generate\n" ] }, { "data": { "text/html": [ "\n", " \n", " " ], "text/plain": [ "" ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" } ], "source": [ "row = df.sample(1).iloc[0]\n", "print(row['text'])\n", "AudioSegment.from_wav(row['audio_filepath'])" ] }, { "cell_type": "code", "execution_count": 71, "id": "70ec1cb9", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "1101it [00:00, 5538.62it/s]\n" ] } ], "source": [ "tot_dur_h = 2\n", "n_space_ms = 500\n", "\n", "info = mediainfo(df.iloc[0]['audio_filepath'])\n", "sample_rate = int(info['sample_rate'])\n", "space_arr = np.zeros(int(round(sample_rate * n_space_ms / 1_000)), dtype=np.int16)\n", "\n", "def _frames_to_ms(arr_len):\n", " return int(round(arr_len / sample_rate * 1_000))\n", "\n", "# calculate number of samples we need\n", "n_head = 0\n", "sum_dur_h = 0\n", "for _, row in df.sample(frac=1.0, random_state=6006).iterrows():\n", " n_head += 1\n", " sum_dur_h += row['duration'] / 60 / 60\n", " if sum_dur_h >= tot_dur_h:\n", " break\n", "\n", "data = []\n", "meta = []\n", "offset_frames = 0\n", "for _, row in tqdm.tqdm(df.sample(frac=1.0, random_state=6006).head(n_head).iterrows()):\n", " audio_segment = AudioSegment.from_wav(row['audio_filepath'])\n", " assert(audio_segment.frame_rate == sample_rate)\n", " assert(audio_segment.channels == 1)\n", " segid = re.sub(r\".+ljspeech\\/LJSpeech\\-1\\.1\\/\", \"\", row['audio_filepath'])\n", " segment_arr = np.array(audio_segment.get_array_of_samples())\n", " \n", " start_ms = _frames_to_ms(offset_frames)\n", " end_ms = _frames_to_ms(offset_frames + segment_arr.shape[0])\n", " offset_frames += segment_arr.shape[0] + space_arr.shape[0]\n", " \n", " data.append(segment_arr)\n", " data.append(space_arr)\n", "\n", " meta.append({\n", " \"start_ms\": start_ms,\n", " \"end_ms\": end_ms,\n", " \"segid\": segid,\n", " })\n", " \n", "new_audio = np.hstack(data)" ] }, { "cell_type": "code", "execution_count": 80, "id": "ea48382c", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "2 hours of audio\n" ] }, { "data": { "text/plain": [ "[{'start_ms': 0, 'end_ms': 9876, 'segid': 'wavs/LJ037-0160.wav'},\n", " {'start_ms': 10376, 'end_ms': 16815, 'segid': 'wavs/LJ014-0071.wav'},\n", " {'start_ms': 17315, 'end_ms': 22941, 'segid': 'wavs/LJ050-0220.wav'}]" ] }, "execution_count": 80, "metadata": {}, "output_type": "execute_result" } ], "source": [ "print(round(_frames_to_ms(new_audio.shape[0] / 1_000 / 60 / 60), 1), \"hours of audio\")\n", "meta[:3]" ] }, { "cell_type": "code", "execution_count": 74, "id": "75f9659e", "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", " \n", " " ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "play_array(new_audio[:sample_rate * 60], sr=sample_rate)" ] }, { "cell_type": "code", "execution_count": 75, "id": "343de129", "metadata": {}, "outputs": [], "source": [ "wavfile.write(\"tmp/ljspeech.wav\", sample_rate, new_audio)" ] }, { "cell_type": "code", "execution_count": 79, "id": "836d8907", "metadata": {}, "outputs": [], "source": [ "with open(\"tmp/ljspeech.json\", \"w\") as f:\n", " json.dump(meta, f)" ] }, { "cell_type": "code", "execution_count": 82, "id": "9c7cfef2", "metadata": {}, "outputs": [], "source": [ "# !ffmpeg -i tmp/ljspeech.wav tmp/ljspeech.mp3" ] }, { "cell_type": "code", "execution_count": null, "id": "5c7bb0dd", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "27336874", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "3d13ca8a", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "c8a997e9", "metadata": {}, "source": [ "## VCTK" ] }, { "cell_type": "code", "execution_count": 85, "id": "97e6aa84", "metadata": {}, "outputs": [], "source": [ "MAIN_DIR = \"/mnt/data-ssd-1/data/vctk/\"" ] }, { "cell_type": "code", "execution_count": 197, "id": "e7b0cb63", "metadata": {}, "outputs": [], "source": [ "# !ls /mnt/data-ssd-1/data/vctk" ] }, { "cell_type": "code", "execution_count": 196, "id": "bad87bbb", "metadata": {}, "outputs": [], "source": [ "# !head -5 /mnt/data-ssd-1/data/vctk/speaker-info.txt" ] }, { "cell_type": "code", "execution_count": 150, "id": "bfa05f9f", "metadata": {}, "outputs": [], "source": [ "# import glob\n", "\n", "# data_txt = {}\n", "# for fn in glob.glob(MAIN_DIR + \"txt/**/*.txt\", recursive=True):\n", "# _id = fn.split(\"/\")[-1][:-4]\n", "# data_txt[_id] = fn\n", " \n", "# data_audio = {}\n", "# for fn in glob.glob(MAIN_DIR + \"wav48_silence_trimmed/**/*.flac\", recursive=True):\n", "# fn_short = fn.split(\"/\")[-1][:-5]\n", "# mic_id = fn_short[-4:]\n", "# _id = fn_short[:-5]\n", "# if _id not in data_audio:\n", "# data_audio[_id] = {}\n", "# data_audio[_id][mic_id] = fn" ] }, { "cell_type": "code", "execution_count": 198, "id": "89c28c58", "metadata": {}, "outputs": [], "source": [ "# pd.Series(data_txt).head(2)" ] }, { "cell_type": "code", "execution_count": 199, "id": "7928d288", "metadata": {}, "outputs": [], "source": [ "# pd.DataFrame(data_audio).T.head(2)" ] }, { "cell_type": "code", "execution_count": 200, "id": "44963ede", "metadata": {}, "outputs": [], "source": [ "# s1 = set(list(data_txt.keys()))\n", "# s2_1 = set([k for k, v in data_audio.items() if \"mic1\" in v])\n", "# s2_2 = set([k for k, v in data_audio.items() if \"mic2\" in v])\n", "\n", "# print(len(s1))\n", "# print(len(s2_1))\n", "# print(len(s2_2))\n", "# print(len(s1 - s2_1 - s2_2))\n", "# print(len((s2_1 | s2_2) - s1))\n", "\n", "# # 44583\n", "# # 44455\n", "# # 43873\n", "# # 300\n", "# # 172" ] }, { "cell_type": "code", "execution_count": 201, "id": "469df8b4", "metadata": {}, "outputs": [], "source": [ "# overlap_set = s1 & s2_1\n", "\n", "# data = []\n", "# for k in tqdm.tqdm(overlap_set):\n", "# text_fp = data_txt[k]\n", "# audio_fp = data_audio[k][\"mic1\"]\n", "# info = mediainfo(audio_fp)\n", "# assert(int(info['sample_rate']) == 48000)\n", "# assert(int(info['channels']) == 1)\n", "# dur_s = float(info[\"duration\"])\n", "# data.append((k, dur_s, text_fp, audio_fp))" ] }, { "cell_type": "code", "execution_count": 202, "id": "ac37f692", "metadata": {}, "outputs": [], "source": [ "# df = pd.DataFrame(data, columns=[\"id\", \"duration_s\", \"text_fp\", \"audio_fp\"])\n", "# print(round(df['duration_s'].sum() / 60 / 60, 1), \"hours total\")\n", "# # 41.5 hours total\n", "# df.head(2)" ] }, { "cell_type": "code", "execution_count": 165, "id": "7f717e48", "metadata": {}, "outputs": [], "source": [ "# df.to_csv(MAIN_DIR + \"custom_manifest.csv\")" ] }, { "cell_type": "code", "execution_count": 203, "id": "fc2121de", "metadata": {}, "outputs": [], "source": [ "df = pd.read_csv(MAIN_DIR + \"custom_manifest.csv\")" ] }, { "cell_type": "code", "execution_count": 204, "id": "72e74108", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ " 20%|██ | 7803/38459 [51:42<3:35:23, 2.37it/s]IOPub message rate exceeded.\n", "The notebook server will temporarily stop sending output\n", "to the client in order to avoid crashing it.\n", "To change this limit, set the config variable\n", "`--NotebookApp.iopub_msg_rate_limit`.\n", "\n", "Current values:\n", "NotebookApp.iopub_msg_rate_limit=1000.0 (msgs/sec)\n", "NotebookApp.rate_limit_window=3.0 (secs)\n", "\n", " 72%|███████▏ | 27782/38459 [3:32:50<1:36:37, 1.84it/s]IOPub message rate exceeded.\n", "The notebook server will temporarily stop sending output\n", "to the client in order to avoid crashing it.\n", "To change this limit, set the config variable\n", "`--NotebookApp.iopub_msg_rate_limit`.\n", "\n", "Current values:\n", "NotebookApp.iopub_msg_rate_limit=1000.0 (msgs/sec)\n", "NotebookApp.rate_limit_window=3.0 (secs)\n", "\n" ] } ], "source": [ "tot_dur_h = 36 # 2\n", "n_space_ms = 500\n", "\n", "sample_rate = 48000\n", "space_arr = np.zeros(int(round(sample_rate * n_space_ms / 1_000)), dtype=np.int16)\n", "\n", "def _frames_to_ms(arr_len):\n", " return int(round(arr_len / sample_rate * 1_000))\n", "\n", "# calculate number of samples we need\n", "n_head = 0\n", "sum_dur_h = 0\n", "for _, row in df.sample(frac=1.0, random_state=6006).iterrows():\n", " n_head += 1\n", " sum_dur_h += row['duration_s'] / 60 / 60\n", " if sum_dur_h >= tot_dur_h:\n", " break\n", "\n", "data = []\n", "meta = []\n", "offset_frames = 0\n", "for _, row in tqdm.tqdm(df.sample(frac=1.0, random_state=6006).head(n_head).iterrows(), total=n_head):\n", " audio_segment = AudioSegment.from_file(row['audio_fp'], codec='flac')\n", " assert(audio_segment.frame_rate == sample_rate)\n", " assert(audio_segment.channels == 1)\n", " segment_arr = np.array(audio_segment.get_array_of_samples())\n", " \n", " start_ms = _frames_to_ms(offset_frames)\n", " end_ms = _frames_to_ms(offset_frames + segment_arr.shape[0])\n", " offset_frames += segment_arr.shape[0] + space_arr.shape[0]\n", " \n", " data.append(segment_arr)\n", " data.append(space_arr)\n", "\n", " meta.append({\n", " \"start_ms\": start_ms,\n", " \"end_ms\": end_ms,\n", " \"segid\": row[\"id\"],\n", " })\n", " \n", "new_audio = np.hstack(data)" ] }, { "cell_type": "code", "execution_count": 205, "id": "f79b6b4d", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "41.3 hours of audio\n" ] }, { "data": { "text/plain": [ "[{'start_ms': 0, 'end_ms': 3066, 'segid': 'p262_071'},\n", " {'start_ms': 3566, 'end_ms': 7382, 'segid': 'p305_044'},\n", " {'start_ms': 7882, 'end_ms': 10656, 'segid': 'p304_153'}]" ] }, "execution_count": 205, "metadata": {}, "output_type": "execute_result" } ], "source": [ "print(round(_frames_to_ms(new_audio.shape[0]) / 1_000 / 60 / 60, 1), \"hours of audio\")\n", "meta[:3]" ] }, { "cell_type": "code", "execution_count": 206, "id": "a6902e85", "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", " \n", " " ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "play_array(new_audio[:sample_rate * 60], sr=sample_rate)" ] }, { "cell_type": "code", "execution_count": null, "id": "373f0d17", "metadata": {}, "outputs": [], "source": [ "new_audio.shape" ] }, { "cell_type": "code", "execution_count": 212, "id": "27b15655", "metadata": {}, "outputs": [], "source": [ "p_len = 2_000_000_000\n", "for n in range(new_audio.shape[0] // p_len + 1):\n", " wavfile.write(\"tmp/vctk_p{}.wav\".format(n + 1), sample_rate, new_audio[n * p_len:(n + 1) * p_len])\n", "print(n + 1, \"parts\")" ] }, { "cell_type": "code", "execution_count": 217, "id": "7cad56d4", "metadata": {}, "outputs": [], "source": [ "with open(\"tmp/vctk.json\", \"w\") as f:\n", " json.dump(meta, f)" ] }, { "cell_type": "code", "execution_count": 222, "id": "6656fa70", "metadata": {}, "outputs": [], "source": [ "# !ffmpeg -i tmp/vctk_p1.wav tmp/vctk_p1.mp3" ] }, { "cell_type": "code", "execution_count": 224, "id": "8dd41b89", "metadata": {}, "outputs": [], "source": [ "# !ffmpeg -i tmp/vctk_p2.wav tmp/vctk_p2.mp3" ] }, { "cell_type": "code", "execution_count": 220, "id": "c715d508", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "ffmpeg version 4.2.4-1ubuntu0.1 Copyright (c) 2000-2020 the FFmpeg developers\n", " built with gcc 9 (Ubuntu 9.3.0-10ubuntu2)\n", " configuration: --prefix=/usr --extra-version=1ubuntu0.1 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --arch=amd64 --enable-gpl --disable-stripping --enable-avresample --disable-filter=resample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libcodec2 --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libjack --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librsvg --enable-librubberband --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-lv2 --enable-omx --enable-openal --enable-opencl --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-nvenc --enable-chromaprint --enable-frei0r --enable-libx264 --enable-shared\n", " libavutil 56. 31.100 / 56. 31.100\n", " libavcodec 58. 54.100 / 58. 54.100\n", " libavformat 58. 29.100 / 58. 29.100\n", " libavdevice 58. 8.100 / 58. 8.100\n", " libavfilter 7. 57.100 / 7. 57.100\n", " libavresample 4. 0. 0 / 4. 0. 0\n", " libswscale 5. 5.100 / 5. 5.100\n", " libswresample 3. 5.100 / 3. 5.100\n", " libpostproc 55. 5.100 / 55. 5.100\n", "\u001b[0;33mGuessed Channel Layout for Input Stream #0.0 : mono\n", "\u001b[0mInput #0, wav, from 'tmp/vctk_p3.wav':\n", " Duration: 11:34:26.67, bitrate: 768 kb/s\n", " Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 48000 Hz, mono, s16, 768 kb/s\n", "Stream mapping:\n", " Stream #0:0 -> #0:0 (pcm_s16le (native) -> mp3 (libmp3lame))\n", "Press [q] to stop, [?] for help\n", "Output #0, mp3, to 'tmp/vctk_p3.mp3':\n", " Metadata:\n", " TSSE : Lavf58.29.100\n", " Stream #0:0: Audio: mp3 (libmp3lame), 48000 Hz, mono, s16p\n", " Metadata:\n", " encoder : Lavc58.54.100 libmp3lame\n", "size= 325521kB time=11:34:26.68 bitrate= 64.0kbits/s speed= 165x \n", "video:0kB audio:325521kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.000071%\n" ] } ], "source": [ "!ffmpeg -i tmp/vctk_p3.wav tmp/vctk_p3.mp3" ] }, { "cell_type": "code", "execution_count": 225, "id": "e4675560", "metadata": {}, "outputs": [], "source": [ "# !ffmpeg -i tmp/vctk_p4.wav tmp/vctk_p4.mp3" ] }, { "cell_type": "code", "execution_count": 229, "id": "2a83acef", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\r\n", "Input File : 'tmp/vctk_p1.mp3'\r\n", "Channels : 1\r\n", "Sample Rate : 48000\r\n", "Precision : 16-bit\r\n", "Duration : 11:34:26.74 = 2000003328 samples ~ 3.12501e+06 CDDA sectors\r\n", "File Size : 333M\r\n", "Bit Rate : 64.0k\r\n", "Sample Encoding: MPEG audio (layer I, II or III)\r\n", "\r\n" ] } ], "source": [ "!soxi tmp/vctk_p1.mp3" ] }, { "cell_type": "code", "execution_count": 234, "id": "22201dd6", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\r\n", "Input File : 'tmp/vctk.mp3'\r\n", "Channels : 1\r\n", "Sample Rate : 48000\r\n", "Precision : 16-bit\r\n", "Duration : 41:20:30.62 = 7143869952 samples ~ 1.11623e+07 CDDA sectors\r\n", "File Size : 1.19G\r\n", "Bit Rate : 64.0k\r\n", "Sample Encoding: MPEG audio (layer I, II or III)\r\n", "\r\n" ] } ], "source": [ "!soxi tmp/vctk.mp3" ] }, { "cell_type": "code", "execution_count": 233, "id": "f4d79b5d", "metadata": {}, "outputs": [], "source": [ "!sox tmp/vctk_p1.mp3 tmp/vctk_p2.mp3 tmp/vctk_p3.mp3 tmp/vctk_p4.mp3 tmp/vctk.mp3" ] }, { "cell_type": "code", "execution_count": 235, "id": "8d6d3224", "metadata": {}, "outputs": [], "source": [ "!rm tmp/vctk_p1.wav tmp/vctk_p2.wav tmp/vctk_p3.wav tmp/vctk_p4.wav" ] }, { "cell_type": "code", "execution_count": 236, "id": "ef0d404b", "metadata": {}, "outputs": [], "source": [ "!rm tmp/vctk_p1.mp3 tmp/vctk_p2.mp3 tmp/vctk_p3.mp3 tmp/vctk_p4.mp3" ] }, { "cell_type": "code", "execution_count": null, "id": "da0292c7", "metadata": {}, "outputs": [], "source": [ "!ls -lah tmp" ] }, { "cell_type": "code", "execution_count": null, "id": "55998291", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "b4ff47a4", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "markdown", "id": "df000979", "metadata": {}, "source": [ "### find specific stuff" ] }, { "cell_type": "code", "execution_count": 174, "id": "f0fc8d31", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'start_ms': 2797953, 'end_ms': 2801027, 'segid': 'p264_253'}" ] }, "execution_count": 174, "metadata": {}, "output_type": "execute_result" } ], "source": [ "meta[726]" ] }, { "cell_type": "code", "execution_count": 176, "id": "26ad7d79", "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", " \n", " " ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "play_array(new_audio[\n", " int(sample_rate * meta[726]['start_ms'] / 1_000) : \n", " int(sample_rate * meta[726]['end_ms'] / 1_000)\n", "], sr=sample_rate)" ] }, { "cell_type": "code", "execution_count": 179, "id": "04767c86", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "He is keen to stay in public service.\n", "\n" ] } ], "source": [ "with open(df[df['id'] == \"p264_253\"].iloc[0]['text_fp']) as f:\n", " print(f.read())" ] }, { "cell_type": "code", "execution_count": 180, "id": "857663f5", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[{'start_ms': 0, 'end_ms': 3066, 'segid': 'p262_071'},\n", " {'start_ms': 3566, 'end_ms': 7382, 'segid': 'p305_044'},\n", " {'start_ms': 7882, 'end_ms': 10656, 'segid': 'p304_153'}]" ] }, "execution_count": 180, "metadata": {}, "output_type": "execute_result" } ], "source": [ "meta[:3]" ] }, { "cell_type": "code", "execution_count": 190, "id": "c60e6c03", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ " 3%|▎ | 70/2138 [00:00<00:00, 18014.56it/s]" ] }, { "name": "stdout", "output_type": "stream", "text": [ "{'start_ms': 265007, 'end_ms': 269112, 'segid': 'p233_294'}\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "\n" ] } ], "source": [ "text = 'family will ever get over this'\n", "\n", "found_id = \"\"\n", "for _, row in tqdm.tqdm(df.sample(frac=1.0, random_state=6006).head(n_head).iterrows(), total=n_head):\n", " with open(row[\"text_fp\"]) as f:\n", " fulltext = f.read()\n", " if text in fulltext:\n", " found_id = row['id']\n", " break\n", " \n", "for m in meta:\n", " if m['segid'] == found_id:\n", " print(m)\n", " break" ] }, { "cell_type": "code", "execution_count": null, "id": "0c88282e", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "79961a6f", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "74354512", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "51959167", "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": "3.8.10" } }, "nbformat": 4, "nbformat_minor": 5 }