audio audioduration (s) 2.32 28.7 | text stringlengths 46 476 | text_ts stringlengths 72 640 | preconditioning stringclasses 9
values | start_time stringlengths 12 12 | end_time stringlengths 12 12 | speech_duration float32 2.32 28.7 | word_timestamps stringlengths 665 5.86k | source_file stringclasses 2
values | language stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|
Finish approach, good evening American 1479, record. American 1479, beach approach, expect 25L. 25L. American 1479, 26 available tonight if you like. We will take 26. American 1479, verify that's you. That is American 1479, sorry, we will take 26. American 1479, expect 26, fly present heading, maintain 5000. Present he... | <|0.00|>Finish approach, good evening American 1479, record.<|2.76|><|2.76|>American 1479, beach approach, expect 25L. 25L.<|4.88|><|4.88|>American 1479, 26 available tonight if you like.<|7.94|><|7.94|>We will take 26.<|10.46|><|10.46|>American 1479, verify that's you.<|12.20|><|12.20|>That is American 1479, sorry,<|1... | 00:00:02.198 | 00:00:29.844 | 27.646 | [{"word": "Finish", "start": 0.10000000000000009, "end": 0.19200000000000017}, {"word": "approach,", "start": 0.2320000000000002, "end": 1.8219999999999996}, {"word": "good", "start": 1.863, "end": 1.9629999999999996}, {"word": "evening", "start": 2.084, "end": 2.305}, {"word": "American", "start": 2.3459999999999996, ... | youtube-atc-2.mp3 | en | ||
Call sign again, fly present heading, maintain 5000. Alright, American 1479, present heading 5000, we'll expect 26. Thank you. American 1479, Scarborough 10 o'clock, 14 miles, advising you to get in sight, lower in just a minute. Alright, it's in sight, American 1479. American 1479, turn left, heading 2-9-0, intercept ... | <|0.00|>Call sign again, fly present heading, maintain 5000.<|2.86|><|2.86|>Alright, American 1479, present<|6.58|><|6.58|>heading 5000, we'll expect 26.<|8.64|><|8.64|>Thank you. American 1479, Scarborough 10 o'clock, 14 miles, advising you<|12.18|><|12.18|>to get in sight, lower in just a minute. Alright, it's in sig... | 00:00:30.167 | 00:00:54.109 | 23.941999 | [{"word": "Call", "start": 0.10000000000000142, "end": 0.2840000000000025}, {"word": "sign", "start": 0.2970000000000006, "end": 0.2970000000000006}, {"word": "again,", "start": 0.34800000000000253, "end": 0.5280000000000022}, {"word": "fly", "start": 0.6490000000000009, "end": 0.7690000000000019}, {"word": "present", ... | youtube-atc-2.mp3 | en | ||
heading, maintain 4000, intercept 26 localizer, American. 1479. American 1479, cleared visual approach runway 26. I have a visual approach, American 1479. American 1479, cleared visual approach runway 26. We've got a communication problem. You're hearing this radio clear because I am saying what I need to say, but some... | <|0.00|>heading, maintain 4000, intercept 26 localizer, American.<|3.12|><|3.12|>1479.<|5.62|><|5.62|>American 1479, cleared visual approach runway 26.<|8.74|><|8.74|>I have a visual approach, American 1479.<|11.38|><|11.38|>American 1479, cleared visual<|14.96|><|14.96|>approach runway 26. We've got a communication<|1... | Call sign again, fly present heading, maintain 5000. Alright, American 1479, present heading 5000, we'll expect 26. Thank you. American 1479, Scarborough 10 o'clock, 14 miles, advising you to get in sight, lower in just a minute. Alright, it's in sight, American 1479. American 1479, turn left, heading 2-9-0, intercept ... | 00:00:56.246 | 00:01:21.014 | 24.768999 | [{"word": "heading,", "start": 0.10000000000000142, "end": 0.28000000000000824}, {"word": "maintain", "start": 0.3200000000000074, "end": 0.6010000000000062}, {"word": "4000,", "start": 0.6210000000000022, "end": 0.7010000000000076}, {"word": "intercept", "start": 0.8420000000000059, "end": 1.122000000000007}, {"word":... | youtube-atc-2.mp3 | en | |
American 1479, you said cleared the approach. I need the runway assignment, which is runway 26, not 25 left or 25 right. You're cleared visual approach runway 26. and you're not reading it back correctly. Clear for the visual approach to runway 26, | <|0.00|>American 1479, you said cleared the approach.<|2.56|><|2.56|>I need the runway assignment, which is runway 26, not 25 left<|6.14|><|6.14|>or 25 right. You're cleared visual approach runway 26. and you're not reading<|9.88|><|9.88|>it back correctly. Clear for the visual approach<|13.56|><|13.56|>to runway 26,<|... | heading, maintain 4000, intercept 26 localizer, American. 1479. American 1479, cleared visual approach runway 26. I have a visual approach, American 1479. American 1479, cleared visual approach runway 26. We've got a communication problem. You're hearing this radio clear because I am saying what I need to say, but some... | 00:01:21.014 | 00:01:36.648 | 15.633 | [{"word": "American", "start": 0.4125000000000085, "end": 0.6535000000000082}, {"word": "1479,", "start": 0.6744999999999948, "end": 0.714500000000001}, {"word": "you", "start": 1.3785000000000025, "end": 1.4185000000000088}, {"word": "said", "start": 1.4784999999999968, "end": 1.6304999999999978}, {"word": "cleared", ... | youtube-atc-2.mp3 | en | |
American 1479. American 1479, if you need a phone number, you can call in, and we can talk about the taste, but you misread the last three transmissions without a call sign and without a runway assignment. | <|0.00|>American 1479. American 1479, if you need a phone number, you can call in, and we can talk about the taste,<|3.74|><|3.74|>but you misread the last three transmissions without a call sign and<|7.24|><|7.24|>without a runway assignment.<|8.26|> | American 1479, you said cleared the approach. I need the runway assignment, which is runway 26, not 25 left or 25 right. You're cleared visual approach runway 26. and you're not reading it back correctly. Clear for the visual approach to runway 26, | 00:01:39.397 | 00:01:47.766 | 8.369 | [{"word": "American", "start": 0.09999999999999432, "end": 0.17099999999999227}, {"word": "1479.", "start": 0.17099999999999227, "end": 0.17099999999999227}, {"word": "American", "start": 0.17099999999999227, "end": 0.465999999999994}, {"word": "1479,", "start": 0.48699999999999477, "end": 0.5679999999999978}, {"word":... | youtube-atc-2.mp3 | en | |
American 1479, do you need a phone number? I'm | <|0.00|>American 1479, do you need a phone number?<|1.88|><|1.88|>I'm<|3.50|> | American 1479. American 1479, if you need a phone number, you can call in, and we can talk about the taste, but you misread the last three transmissions without a call sign and without a runway assignment. | 00:01:49.717 | 00:01:53.311 | 3.594 | [{"word": "American", "start": 0.09999999999999432, "end": 0.36199999999999477}, {"word": "1479,", "start": 0.382000000000005, "end": 0.402000000000001}, {"word": "do", "start": 0.4129999999999967, "end": 0.4129999999999967}, {"word": "you", "start": 0.4129999999999967, "end": 0.4129999999999967}, {"word": "need", "sta... | youtube-atc-2.mp3 | en | |
not going to waste my time. American 1479, contact tower, 118.7. 118.7 American 1479 | <|0.00|>not going to waste my time. American<|3.66|><|3.66|>1479, contact tower, 118.7. 118.7 American 1479<|5.72|> | American 1479, do you need a phone number? I'm | 00:01:55.639 | 00:02:01.455 | 5.816 | [{"word": "not", "start": 0.09999999999999432, "end": 0.21999999999998465}, {"word": "going", "start": 0.2799999999999869, "end": 0.5009999999999906}, {"word": "to", "start": 0.5809999999999889, "end": 0.6319999999999908}, {"word": "waste", "start": 0.8629999999999853, "end": 1.043999999999997}, {"word": "my", "start":... | youtube-atc-2.mp3 | en | |
Winter Haven traffic, 455 Uniform Sarah is a mile north of the airport on a midfield crosswind to get to a left downwind for runway 29. We're looking for the guy on 5. We're just shooting a practice approach and we'll be a touch and go from 5. 455 Uniform Sarah, we're looking. We're just on top of the airport right now... | <|0.00|>Winter Haven traffic, 455 Uniform Sarah is a mile north<|3.76|><|3.76|>of the airport on a midfield crosswind to get to a left downwind<|7.62|><|7.62|>for runway 29. We're looking for the guy on 5.<|11.64|><|11.64|>We're just shooting a practice approach and we'll be a<|15.16|><|15.16|>touch and go from<|18.84|... | 00:00:07.732 | 00:00:36.429 | 28.697001 | [{"word": "Winter", "start": 0.09999999999999964, "end": 0.2919999999999989}, {"word": "Haven", "start": 0.512999999999999, "end": 0.6139999999999999}, {"word": "traffic,", "start": 0.7549999999999999, "end": 1.1569999999999991}, {"word": "455", "start": 1.1770000000000005, "end": 1.258}, {"word": "Uniform", "start": 2... | youtube-atc-1.m4a | en | ||
right, we're on a two-mile final for five. Eight-fines to your papa. Winter Haven traffic, tricky. Eight-fines to your papa on a short final for runway five. We'll be a touch and go. Winter Haven. Yeah, I see that. I was just about to go miss, so I'm going to miss. Winter Haven traffic, tricky. Eight-fines to your papa... | <|0.00|>right, we're on a two-mile<|3.60|><|3.60|>final for five. Eight-fines to your papa.<|5.34|><|5.34|>Winter Haven traffic, tricky. Eight-fines to your papa on a short final for runway<|9.16|><|9.16|>five. We'll be a touch and go. Winter Haven.<|11.64|><|11.64|>Yeah, I see that. I was just about to go miss, so I'm... | Winter Haven traffic, 455 Uniform Sarah is a mile north of the airport on a midfield crosswind to get to a left downwind for runway 29. We're looking for the guy on 5. We're just shooting a practice approach and we'll be a touch and go from 5. 455 Uniform Sarah, we're looking. We're just on top of the airport right now... | 00:00:38.356 | 00:01:05.709 | 27.353001 | [{"word": "right,", "start": 0.10000000000000142, "end": 0.2609999999999957}, {"word": "we're", "start": 2.0259999999999962, "end": 2.1469999999999985}, {"word": "on", "start": 2.5079999999999956, "end": 2.5279999999999987}, {"word": "a", "start": 3.370999999999995, "end": 3.370999999999995}, {"word": "two-mile", "star... | youtube-atc-1.m4a | en | |
Oh, nice. How about if you just shut the fuck up if you're in a pickup truck and ain't worried about flying an airplane then, dickhead? Kick rocks, man. | <|0.00|>Oh, nice. How about if you just shut the fuck up if<|3.54|><|3.54|>you're in a pickup truck and ain't worried about flying an airplane then, dickhead?<|6.90|><|6.90|>Kick rocks, man.<|9.64|> | right, we're on a two-mile final for five. Eight-fines to your papa. Winter Haven traffic, tricky. Eight-fines to your papa on a short final for runway five. We'll be a touch and go. Winter Haven. Yeah, I see that. I was just about to go miss, so I'm going to miss. Winter Haven traffic, tricky. Eight-fines to your papa... | 00:01:07.678 | 00:01:17.423 | 9.745 | [{"word": "Oh,", "start": 0.09999999999999432, "end": 0.11999999999999034}, {"word": "nice.", "start": 0.18099999999998317, "end": 0.340999999999994}, {"word": "How", "start": 1.8279999999999887, "end": 1.907999999999987}, {"word": "about", "start": 2.008999999999986, "end": 2.1799999999999926}, {"word": "if", "start":... | youtube-atc-1.m4a | en | |
You obviously don't need to be flying an airplane if you have an attitude like | <|0.00|>You obviously don't need to be flying an airplane if you have an attitude like<|2.40|> | Oh, nice. How about if you just shut the fuck up if you're in a pickup truck and ain't worried about flying an airplane then, dickhead? Kick rocks, man. | 00:01:21.366 | 00:01:23.873 | 2.507 | [{"word": "You", "start": 0.09999999999999432, "end": 0.17300000000000182}, {"word": "obviously", "start": 0.21399999999999864, "end": 0.4170000000000016}, {"word": "don't", "start": 0.49800000000000466, "end": 0.5990000000000038}, {"word": "need", "start": 0.6400000000000006, "end": 0.7409999999999997}, {"word": "to",... | youtube-atc-1.m4a | en | |
check the APAC here. I'll put the patterns to the north. | <|0.00|>check the APAC here. I'll put the patterns to the north.<|2.22|> | 00:01:24.829 | 00:01:27.151 | 2.322 | [{"word": "check", "start": 0.09999999999999432, "end": 0.2219999999999942}, {"word": "the", "start": 0.262999999999991, "end": 0.3229999999999933}, {"word": "APAC", "start": 0.6609999999999872, "end": 0.6609999999999872}, {"word": "here.", "start": 0.7009999999999934, "end": 0.833999999999989}, {"word": "I'll", "start... | youtube-atc-1.m4a | en |
atc-draft-with-confidence-filter
Speech dataset prepared with Trelis Studio.
Statistics
| Metric | Value |
|---|---|
| Source files | 2 |
| Train samples | 12 |
| Total duration | 3.5 minutes |
Columns
| Column | Type | Description |
|---|---|---|
audio |
Audio | Audio segment (16kHz) - speech only, extracted from aligned regions |
text |
string | Plain transcription (no timestamps) |
text_ts |
string | Transcription with Whisper timestamp tokens |
preconditioning |
string | Previous segment's plain text (empty for first segment of each source file) |
start_time |
string | Segment start in original audio (HH:MM:SS.mmm) |
end_time |
string | Segment end in original audio (HH:MM:SS.mmm) |
speech_duration |
float | Duration of speech in segment (excluding silence) |
word_timestamps |
string | Word-level timestamps as JSON (relative to speech-only audio) |
source_file |
string | Original audio filename |
language |
string | ISO 639-1 language code for Whisper language token (e.g., en, el) |
Speech Segmentation
For .txt transcripts, CTC alignment produces word-level timestamps that define speech
boundaries. The full audio is treated as a single speech region and split at word boundaries
into ~20s chunks. For .srt/.vtt files, transcript timestamps define segments.
Timestamps are relative to the extracted speech audio within each chunk.
Training Usage
2-bucket approach (default)
- Bucket A (50%): Use
textcolumn - plain transcription without timestamps - Bucket B (50%): Use
text_tscolumn - transcription with Whisper timestamp tokens
4-bucket approach (with preconditioning)
- Bucket A (25%):
textonly (no timestamps, no preconditioning) - Bucket A' (25%):
preconditioning+text(previous segment context, no timestamps) - Bucket B (25%):
text_tsonly (timestamps, no preconditioning) - Bucket B' (25%):
preconditioning+text_ts(previous segment context + timestamps)
Preconditioning prepends the previous segment's text to teach the model conversational continuity. Bucket ratios are configurable at training time.
Usage
from datasets import load_dataset
dataset = load_dataset("Trelis/atc-draft-with-confidence-filter")
Prepared with Trelis Studio
- Downloads last month
- 17