Dataset Viewer
Auto-converted to Parquet Duplicate
idx
int32
0
10k
split
stringclasses
1 value
image
imagewidth (px)
150
640
audio
audioduration (s)
1.42
21.4
long_desc
stringlengths
428
1.65k
sst
stringlengths
1
150
processed_sst
stringlengths
1
150
audio_tokens_mimi
listlengths
8
8
audio_tokens_encodec
listlengths
16
16
5,791
train
The image depicts a large indoor parking area filled with numerous motorcycles. The motorcycles are parked in rows, creating a dense and organized arrangement. The floor is tiled with a pattern that includes a combination of beige and brown tiles, giving the space a clean and modern look. In the foreground, there are ...
A large indoor motorcycle parking area with rows of bikes, bright lighting, and clean floors.
A large indoor motorcycle parking area with rows of bikes, bright lighting, and clean floors.
[ [ 1049, 127, 1272, 349, 383, 114, 1610, 1741, 828, 1536, 1536, 374, 67, 631, 114, 1610, 468, 1932, 1437, 321, 1141, 1690, 1690, 622, 628, 1011, 1881, 584, 471, 8, 304, 1655, 1624, ...
[ [ 62, 62, 62, 62, 408, 408, 62, 408, 62, 408, 62, 62, 62, 738, 1017, 63, 869, 362, 314, 646, 264, 687, 311, 857, 857, 759, 776, 867, 869, 352, 434, 941, 941, 595, 870, 270,...
343
train
The image depicts a snowy mountain landscape with two skiers standing on a snow-covered slope. The sky is overcast with dense, low-lying clouds, suggesting a cold and possibly windy day. The skiers are positioned on the right side of the image, with one skier slightly ahead of the other. The skier in the foreground i...
Two skiers stand on a snow-covered slope at an elevation likely above tree line, surrounded by a rugged mountain range with dense, low-lying clouds.
Two skiers stand on a snow-covered slope at an elevation likely above tree line, surrounded by a rugged mountain range with dense, low-lying clouds.
[ [ 1049, 127, 1156, 1802, 1274, 1532, 1927, 785, 777, 1395, 1443, 1751, 201, 776, 264, 544, 1902, 80, 1967, 1161, 200, 1399, 1697, 968, 615, 1916, 1135, 1428, 1344, 304, 260, 1593, 80, ...
[ [ 62, 62, 62, 62, 408, 408, 62, 408, 62, 408, 62, 62, 62, 408, 835, 574, 125, 148, 488, 276, 219, 816, 924, 657, 840, 943, 234, 704, 70, 934, 47, 47, 160, 47, 160, 47, ...
7,126
train
The image depicts a scene at an airport during sunset. The primary object in the image is a large commercial airplane, specifically a United Airlines jet, which is parked on the tarmac. The airplane is silhouetted against the setting sun, creating a dramatic and visually striking contrast. The sun is low in the sky, ca...
A United Airlines jet parked on the tarmac at an airport during sunset, with the sun casting a warm glow over the scene, adding a sense of.
A United Airlines jet parked on the tarmac at an airport during sunset, with the sun casting a warm glow over the scene, adding a sense of.
[ [ 1049, 127, 2035, 337, 125, 321, 1440, 313, 1184, 67, 561, 937, 1359, 610, 946, 1532, 1506, 1109, 611, 930, 736, 905, 1356, 96, 1333, 442, 1932, 1932, 1293, 1879, 885, 104, 1467, ...
[ [ 62, 62, 62, 62, 408, 408, 62, 408, 62, 408, 62, 62, 62, 62, 62, 585, 535, 522, 419, 407, 151, 368, 465, 407, 727, 924, 477, 924, 813, 80, 722, 476, 938, 215, 270, 462, ...
2,449
train
The image depicts a group of people on a snowy mountain, likely at a ski resort. The scene is lively and colorful, with various individuals engaged in skiing and snowboarding activities. The following objects and elements are visible: 1. **People**: There are five people in the image, all of whom are engaged in skiing...
A lively crowd of five people are skiing and snowboarding on a snowy mountain, enjoying a festive atmosphere typical of a ski resort.
A lively crowd of five people are skiing and snowboarding on a snowy mountain, enjoying a festive atmosphere typical of a ski resort.
[ [ 1049, 127, 1766, 1513, 1781, 1465, 1214, 1791, 1336, 1336, 1420, 651, 988, 2014, 1721, 1768, 1571, 1615, 6, 1432, 954, 1640, 729, 269, 1945, 200, 224, 553, 49, 796, 247, 358, 1803, ...
[ [ 62, 62, 62, 62, 408, 408, 62, 408, 62, 408, 408, 62, 62, 408, 1017, 600, 377, 333, 252, 687, 858, 944, 784, 467, 57, 539, 775, 775, 562, 270, 55, 230, 908, 81, 503, 945, ...
6,374
train
"The image depicts a cozy, warmly lit kitchen or dining area. The room has a wooden floor with a lig(...TRUNCATED)
"A cozy, warmly lit kitchendining area features a wooden floor, light yellow walls, and a white ceil(...TRUNCATED)
"A cozy, warmly lit kitchendining area features a wooden floor, light yellow walls, and a white ceil(...TRUNCATED)
[[1049,127,1272,510,1705,1782,651,1790,1229,15,577,1962,997,1184,1764,722,862,1003,1065,1817,2046,90(...TRUNCATED)
[[62,62,62,62,408,408,62,408,62,408,62,62,62,408,738,63,989,55,655,1011,1019,408,876,1017,876,785,25(...TRUNCATED)
3,739
train
"The image depicts a scene featuring a bridge over a river, with a train traveling across it. The br(...TRUNCATED)
A bridge over a river features a passenger train traveling across it.
A bridge over a river features a passenger train traveling across it.
[[1049,127,2035,1862,578,1596,264,268,1708,1473,1532,937,1359,957,1416,485,96,442,1624,1231,86,1276,(...TRUNCATED)
[[62,62,62,62,408,408,62,408,62,408,408,62,62,408,1017,336,484,333,578,699,491,321,604,1019,62,785,1(...TRUNCATED)
1,550
train
"The image depicts a dining table with a white tablecloth and a patterned tablecloth underneath. On (...TRUNCATED)
"A dining table with a clear glass containing alcohol and food remains, featuring a red velvet or ch(...TRUNCATED)
"A dining table with a clear glass containing alcohol and food remains, featuring a red velvet or ch(...TRUNCATED)
[[1049,127,2035,1071,181,371,188,1932,266,807,560,1005,1005,1892,13,278,1361,1250,628,510,1252,1344,(...TRUNCATED)
[[62,62,62,62,408,408,62,408,62,408,62,62,62,408,738,433,989,16,692,55,642,687,224,321,491,876,115,7(...TRUNCATED)
2,106
train
"The image depicts a serene savanna landscape with two zebras grazing in the grass. The zebras are t(...TRUNCATED)
"A serene savanna landscape with two zebras grazing in a tall, golden-brown grass field, surrounded (...TRUNCATED)
"A serene savanna landscape with two zebras grazing in a tall, golden-brown grass field, surrounded (...TRUNCATED)
[[1049,127,1156,1802,1063,1790,923,354,1295,1624,1562,109,1670,560,438,912,284,206,1943,514,774,1479(...TRUNCATED)
[[62,62,62,62,408,408,62,408,62,408,408,62,62,408,738,662,989,361,850,349,47,47,160,47,160,160,160,2(...TRUNCATED)
572
train
"The image depicts a collection of fresh fruits arranged on a smooth, light-colored surface, possibl(...TRUNCATED)
"A colorful array of fresh fruits arranged on a tabletop, including a banana with a blue sticker, an(...TRUNCATED)
"A colorful array of fresh fruits arranged on a tabletop, including a banana with a blue sticker, an(...TRUNCATED)
[[1049,127,1272,1862,1705,1003,1580,1580,1840,1517,22,121,1721,1596,263,1976,591,1851,1851,1916,1916(...TRUNCATED)
[[62,62,62,62,408,408,62,408,408,408,408,62,62,62,62,835,146,377,556,255,780,408,121,121,408,63,420,(...TRUNCATED)
4,098
train
"The image depicts a person standing on a wooden floor, holding a skateboard in their left hand. The(...TRUNCATED)
"A man stands on a wooden floor with a white skateboard holding a small skateboard box. He wears dar(...TRUNCATED)
"A man stands on a wooden floor with a white skateboard holding a small skateboard box. He wears dar(...TRUNCATED)
[[1049,127,1272,886,313,159,128,1534,1854,1135,201,776,264,29,641,1644,689,997,1853,1906,1385,858,90(...TRUNCATED)
[[62,62,62,62,408,408,62,408,62,408,62,62,62,62,408,63,989,562,562,687,58,58,58,58,842,202,144,790,7(...TRUNCATED)
End of preview. Expand in Data Studio

VLS 10K

9,987 MS-COCO images paired with a long written description, a one-sentence spoken summary of that description, the spoken audio, and that audio pre-encoded by two neural codecs. Images and audio are embedded in the parquet, so the viewer renders them and one call opens the set:

from datasets import load_dataset
ds = load_dataset("seonglae/vls-10k", split="train")
ds[0]["image"]   # PIL image
ds[0]["audio"]   # decoded waveform
ds[0]["sst"]     # the sentence that was spoken

Columns

column type meaning
idx int32 stable row id
split string train / validation / test
image Image the COCO image
audio Audio speech rendering of sst (wav)
long_desc string long written description of the image
sst string short spoken text, a one-sentence summary of long_desc
processed_sst string sst after text normalisation
audio_tokens_mimi int32[8][T] Mimi codes, 8 codebooks at 12.5 Hz
audio_tokens_encodec int32[16][T] EnCodec codes, 16 codebooks

Both token columns are [codebooks][frames], so their shape carries the frame and codebook counts directly.

Notes

Every row carries two independent tokenizations of the same waveform, which makes the set usable for comparing codecs with the text, image and audio held fixed.

The split ships as a column and as separate parquet files, so it never has to be re-derived and cannot drift between users.

Images come from MS-COCO and keep its terms. The descriptions, spoken summaries and audio are generated.

A larger companion set with 74,936 rows is at seonglae/vls-100k.

Downloads last month
26