Update the dataset card

README.md

@@ -26,24 +26,26 @@ dataset_info:
         dtype: audio
       - name: topics
         dtype: string
+      - name: relationship
+        dtype: string
       - name: duration_min
         dtype: int64
       - name: turns_per_minute
         dtype: float64
-      - name: speaker_a
-        dtype: string
-      - name: speaker_b
-        dtype: string
+      - name: turn_taking_gap_ms
+        dtype: int64
+      - name: speech_dominance
+        dtype: float64
 configs:
   - config_name: audio_previews
     data_files:
-      - split: train
+      - split: Sample
         path: "preview/**"
 ---
 
 # Open Yap 1K sample
 
-**The world's largest free conversational dataset: 1,000 hours of two-speaker English speech, open to labs and research teams worldwide.**
+**The world's largest free conversational dataset: 1000 hours of two-speaker English speech, open to labs and research teams worldwide.**
 
 With this open release, we aim to close a gap in the literature: speech recorded as it
 happens in real life. Fisher and Switchboard assigned partners and topics to maximize
@@ -55,22 +57,25 @@ That design choice shows in the data quality. People who know each other interru
 more, backchannel more, and leave shorter gaps between turns: the behaviour a
 full-duplex model has to learn.
 
-These 1,000 hours are one part of a larger licensed corpus we build with frontier
+These 1000 hours are one part of a larger licensed corpus we build with frontier
 labs and research teams. We release them because progress in conversational AI is
 slower than it needs to be - and open data is the fastest way to change that for
 everyone. Contact us to discuss licensing the full corpus.
 
-The HuggingFace repository only holds a sample. The full corpus is publicly available
+**The HuggingFace repository only holds a sample.** The full corpus is publicly available
 for commercial and research use, under the Open Yap 1K Data Use Agreement. Request it at
-https://theagenticdatacompany.com/open-yap-1k.
+[theagenticdatacompany.com/open-yap-1k](https://theagenticdatacompany.com/open-yap-1k).
+
+## Corpus composition
 
 <table style="display:table;width:100%">
-<thead><tr><th></th><th>HuggingFace</th><th>Full Corpus</th></tr></thead>
+<thead><tr><th style="text-align:left"></th><th style="text-align:left">HuggingFace</th><th style="text-align:left">Full Corpus</th></tr></thead>
 <tbody>
-<tr><td>Size</td><td>16 conversations, 8.9 h</td><td>1,602 conversations, 1,000 h</td></tr>
+<tr><td>Size</td><td>8.9 h, 16 conversations</td><td>1000 h, 1602 conversations</td></tr>
 <tr><td>Speakers</td><td>8</td><td>239</td></tr>
-<tr><td>Licence</td><td>CC-BY-4.0</td><td>Open Yap 1K Data Use Agreement</td></tr>
-<tr><td>Access</td><td>This page</td><td>Request form</td></tr>
+<tr><td>License</td><td>CC-BY-4.0</td><td>Open Yap 1K Data Use Agreement</td></tr>
+<tr><td>Access</td><td>This page</td><td><a href="https://theagenticdatacompany.com/open-yap-1k">Request form</a></td></tr>
+<tr><td>Audio</td><td>48 kHz, 16-bit FLAC, one file per speaker</td><td>48 kHz, 16-bit PCM, one file per speaker</td></tr>
 </tbody>
 </table>
 
@@ -80,42 +85,25 @@ Website: [theagenticdatacompany.com](https://theagenticdatacompany.com)
 
 Email: christian@theagenticdatacompany.com
 
-## Corpus composition
-
-| | |
-|---|---|
-| Conversations | 1,602 (1,136 friends, 173 colleagues, 157 romantic partners, 136 family) |
-| Duration | 1,000 h, averaging 37 min |
-| Speakers | 239, all adults, in 34 countries |
-| Language | English (en) |
-| Audio | 48 kHz, 16-bit PCM, one file per speaker |
-
-## What is in this repository
-
-16 conversations (8.9 hours) drawn from the corpus, twice over:
-
-- **The data: WebDataset shards**, `shard-*.tar`, holding both channels as FLAC,
-  the listening copy and the full record. Download them from the Files tab or
-  with the `hf` CLI, and load them with the WebDataset loader, below.
-- **The table: `audio_previews`**, the one config declared for the Data Studio.
-  One row per conversation with the listening copy and the scannable facts, from
-  `preview/`. The shards are not declared as a config because the Hub's viewer
-  cannot convert tar shards today (an issue on their side); loading them
-  yourself works as it always has.
-
-**The packaging differs from the full delivery**, which ships per-conversation
-directories; the contents are the same. Each shard member:
-
-| member | contents |
-|---|---|
-| `<key>.preview.mp3` | **a listening copy, not the data**: the whole conversation as stereo MP3, speaker A left, speaker B right |
-| `<key>.topics.txt` | subject tags, comma-separated |
-| `<key>.duration_min.json` | conversation length in minutes |
-| `<key>.turns_per_minute.json` | replies and interruptions per minute |
-| `<key>.speaker_a.txt`, `<key>.speaker_b.txt` | age range · gender · country · native or non-native English |
-| `<key>.a.flac` | speaker A's channel — mono, 48 kHz, 16-bit |
-| `<key>.b.flac` | speaker B's channel — same |
-| `<key>.json` | the full record: conversation and per-speaker metadata, and both transcripts |
+## Dataset structure
+
+16 conversations (8.9 hours) drawn from the corpus. One
+sample per conversation, both speakers on one timeline.
+
+```text
+open-yap-1k/
+├── README.md
+├── LICENSE.txt
+├── metadata.jsonl               one line per conversation
+├── preview/
+│   ├── metadata.jsonl           the Data Studio table, one row per conversation
+│   └── <key>.mp3                listening copy: stereo MP3, speaker A left, speaker B right
+└── shard-00000.tar to shard-00001.tar   WebDataset shards, one sample per conversation
+    ├── <key>.preview.mp3        the same listening copy
+    ├── <key>.a.flac             speaker A: mono, 48 kHz, 16-bit
+    ├── <key>.b.flac             speaker B: same timeline, same length
+    └── <key>.json               conversation metadata, both speakers, both transcripts
+```
 
 ```python
 from datasets import load_dataset
@@ -137,12 +125,51 @@ Or fetch the files and read them with any tar reader:
 hf download TheAgenticDataCompany/open-yap-1k --repo-type dataset --include "shard-*.tar" --local-dir open-yap-1k
 ```
 
-`metadata.jsonl` lists every key with its language, relationship, duration,
-topics, and turn rate, so
-the sample can be filtered without opening a shard. Sample 0 of `a.flac` and
-sample 0 of `b.flac` are the same instant — aligning them is a no-op.
-
-## Known limitations
+| File | Metadata it holds |
+|---|---|
+| `preview/metadata.jsonl` | `topics`, `relationship`, `duration_min`, `turns_per_minute`, `turn_taking_gap_ms`, `speech_dominance` |
+| `metadata.jsonl` | `key`, `language`, `relationship`, `topics`, `duration_seconds`, `turns_per_minute`, `turn_taking_gap_ms`, `speech_dominance` |
+| `<key>.json` in each shard | the full record: the conversation fields, `speakers.{a,b}` and `transcripts.{a,b}`, listed under Fields |
+
+## Collection method and quality assurance
+
+We collect the audio with our own app, which works like a phone call. A speaker
+invites someone they know, and the two of them simply talk. This gives us natural
+conversation and a straightforward way to scale collection. The trade-off is less
+control over the recording environment, so real-world noise appears more often. We
+keep the subtle noise, because it teaches a model robustness. However, we remove
+the long tail of noisy recordings with several ML models combined into one pipeline.
+
+|  | Other datasets | Open Yap 1K |
+|---|---|---|
+| Speakers | Strangers, paired by the collector | Friends and family, self-paired |
+| Setting | A treated room | Real rooms, on their own devices |
+| Conversation | An assigned topic | Free talk, any topic |
+| What you hear | Clean turns, little overlap | Overlap, quick turns, backchannel, laughter |
+| Background | Pristine | Slightly noisier, kept on purpose |
+
+### Consent and privacy
+
+- All audio was recorded on our own platform.
+- Speakers register, give explicit consent before their first recording, and are paid for their time.
+- Demographics are self-reported at registration, before any recording, and are never inferred from audio.
+- Speaker identifiers are pseudonymous and stable within the release. Names, contact details and account identifiers are excluded.
+
+Each side is captured on the speaker's own device as uncompressed PCM. The
+recording is separate from the call audio, so no call codec touches the files.
+The two tracks are aligned on one shared timeline and delivered un-normalised, so
+the original dynamics survive. Transcripts come from automatic speech recognition
+with word-level timings.
+
+Before a conversation enters the corpus, a human reviewer listens to it and rates
+language proficiency, naturalness and expressivity. Conversations that sound read
+or performed rather than spontaneous are rejected. For this repository, both
+speakers were then shown the specific recording and agreed, conversation by
+conversation, to publish it. Every published track is measured after review:
+loudness, true peak, noise floor, effective bandwidth, DNSMOS and the silence
+profile ship in `audio_metrics`, so a reader can filter on them.
+
+### Known limitations
 
 - **Hand-picked, not a random draw.** 16 friends;
   nothing about this sample's distribution generalises to the corpus.
@@ -170,6 +197,8 @@ sample 0 of `b.flac` are the same instant — aligning them is a no-op.
 | `duration_seconds` | conversation length |
 | `topics` | lowercase subject tags, each at most three words |
 | `turns_per_minute` | replies and interruptions per complete recording minute, excluding short overlapping backchannels |
+| `turn_taking_gap_ms` | median silence between one speaker stopping and the other starting, in milliseconds |
+| `speech_dominance` | speaker A's share of the transcript words, 0 to 1, where 0.5 is an even split |
 | `sample_rate` | 48000 |
 | `speakers.{a,b}` | per-speaker metadata, below |
 | `transcripts.{a,b}` | that speaker's transcript, below |
@@ -188,21 +217,19 @@ sample 0 of `b.flac` are the same instant — aligning them is a no-op.
 | `device` | capture device or audio-route label |
 | `audio_metrics` | measured from the delivered audio |
 
-`transcripts.{a,b}`: `{ conversation_id, speaker_index, language, text, words[],
-corrections_applied }`, each word `{ word, start, end, type }`. Timings are seconds
-on the shared timeline, `null` only for words a reviewer typed in. Transcription is
-Deepgram Nova-3, word-level, and **not human-verified**.
-
 **Audio is delivered un-normalised.** Integrated loudness and true peak are
 measured and reported so a target level can be applied without probing every file.
 
-## Quality assurance
+`transcripts.{a,b}`:
 
-1. **Human review** — a reviewer approved each conversation in this repository
-   for language proficiency, naturalness and expressivity. Conversations judged
-   read or performed rather than spontaneous are rejected.
-2. **Speaker approval** — both speakers were shown this specific recording and
-   agreed, conversation by conversation, to publish it.
+| field | meaning |
+|---|---|
+| `conversation_id` | the same pseudonymous conversation ID |
+| `speaker_index` | `a` or `b` |
+| `language` | BCP-47 language tag |
+| `text` | the full transcript |
+| `words[]` | one entry per word: `{ word, start, end, type }`. `start` and `end` are seconds on the shared timeline, `null` only for a word a reviewer typed in. `type` is `word`, `filler`, `laugh`, `cough` or `noise` |
+| `corrections_applied` | `true` when a reviewer corrected the machine transcript |
 
 ## Citation
 

metadata.jsonl

@@ -1,16 +1,16 @@
-{"key":"conv_bcff0c2a0be7","language":"en","relationship":"friends","topics":["halloween","horror movies","childhood memories","religion and upbringing","relationships","holiday traditions"],"turns_per_minute":9.763,"duration_seconds":2163.250667}
-{"key":"conv_756ee444d825","language":"en","relationship":"friends","topics":["music and concerts","relationships","communication habits","retail work","personal style","perfume and scent"],"turns_per_minute":9.19,"duration_seconds":4419.807}
-{"key":"conv_d4005da6db98","language":"en","relationship":"friends","topics":["daily routines","travel planning","visa and paperwork","martial arts training","soccer","world cup"],"turns_per_minute":7.165,"duration_seconds":1716.645333}
-{"key":"conv_31bedfc99b53","language":"en","relationship":"friends","topics":["soccer","world cup","sports predictions","gym and fitness","immigration policy","sports organizations"],"turns_per_minute":4.465,"duration_seconds":1384.152}
-{"key":"conv_cb37687ec915","language":"en","relationship":"friends","topics":["food and recipes","weather","relocation","parenting adult children","pets and cats","grief and loss"],"turns_per_minute":14.895,"duration_seconds":2557.88}
-{"key":"conv_0c917fc63b3b","language":"en","relationship":"friends","topics":["career changes","software startups","space travel","aviation","flat earth theory","conspiracy theories"],"turns_per_minute":2.687,"duration_seconds":1183.346}
-{"key":"conv_10cf7f6ff081","language":"en","relationship":"friends","topics":["evolution","human origins","biology","dna and genetics","scientific debate"],"turns_per_minute":3.704,"duration_seconds":1376.882}
-{"key":"conv_e11edcca8521","language":"en","relationship":"friends","topics":["prescription medication","celebrity cosmetic surgery","perfume and fragrances","consumer culture","weather and flooding","online shopping livestreams"],"turns_per_minute":16.147,"duration_seconds":1575.568}
-{"key":"conv_04692967d40e","language":"en","relationship":"friends","topics":["formula one","motorsport","sports viewing","live events","car racing rules"],"turns_per_minute":3.9,"duration_seconds":1015.274667}
-{"key":"conv_5a798c657bb0","language":"en","relationship":"friends","topics":["family relationships","weather","introversion vs extroversion","social encounters","extended family history","health scares"],"turns_per_minute":9.332,"duration_seconds":1273.058667}
-{"key":"conv_2e4aec4666d8","language":"en","relationship":"friends","topics":["football","sports history","athletes","rivalries","world cup"],"turns_per_minute":4.968,"duration_seconds":1002.448}
-{"key":"conv_aa4777220a1e","language":"en","relationship":"friends","topics":["horror movies","1980s nostalgia","first jobs","retail work","music videos","child actors"],"turns_per_minute":7.89,"duration_seconds":3604.53866666667}
-{"key":"conv_b98b64206aee","language":"en","relationship":"friends","topics":["books and reading","self-improvement","artificial intelligence","formula one","motogp","endurance racing"],"turns_per_minute":5.082,"duration_seconds":1015.368}
-{"key":"conv_31982da49c73","language":"en","relationship":"friends","topics":["online shopping","thrifting","fashion","hair color","estranged family","weight loss"],"turns_per_minute":16.296,"duration_seconds":2724.676333}
-{"key":"conv_14f897f75018","language":"en","relationship":"friends","topics":["home appliances","formula one","auto racing training","basketball","nba finals","american sports culture"],"turns_per_minute":4.108,"duration_seconds":1402.070667}
-{"key":"conv_f5c0b02f01e2","language":"en","relationship":"friends","topics":["trivia quizzes","geography","world history","science facts","food and drink","travel plans"],"turns_per_minute":19.77,"duration_seconds":3705.672}
+{"key":"conv_bcff0c2a0be7","language":"en","relationship":"friends","topics":["halloween","horror movies","childhood memories","religion and upbringing","relationships","holiday traditions"],"turns_per_minute":9.763,"turn_taking_gap_ms":400,"speech_dominance":0.3966,"duration_seconds":2163.250667}
+{"key":"conv_756ee444d825","language":"en","relationship":"friends","topics":["music and concerts","relationships","communication habits","retail work","personal style","perfume and scent"],"turns_per_minute":9.19,"turn_taking_gap_ms":340,"speech_dominance":0.3964,"duration_seconds":4419.807}
+{"key":"conv_d4005da6db98","language":"en","relationship":"friends","topics":["daily routines","travel planning","visa and paperwork","martial arts training","soccer","world cup"],"turns_per_minute":7.165,"turn_taking_gap_ms":440,"speech_dominance":0.5058,"duration_seconds":1716.645333}
+{"key":"conv_31bedfc99b53","language":"en","relationship":"friends","topics":["soccer","world cup","sports predictions","gym and fitness","immigration policy","sports organizations"],"turns_per_minute":4.465,"turn_taking_gap_ms":640,"speech_dominance":0.4509,"duration_seconds":1384.152}
+{"key":"conv_cb37687ec915","language":"en","relationship":"friends","topics":["food and recipes","weather","relocation","parenting adult children","pets and cats","grief and loss"],"turns_per_minute":14.895,"turn_taking_gap_ms":440,"speech_dominance":0.4888,"duration_seconds":2557.88}
+{"key":"conv_0c917fc63b3b","language":"en","relationship":"friends","topics":["career changes","software startups","space travel","aviation","flat earth theory","conspiracy theories"],"turns_per_minute":2.687,"turn_taking_gap_ms":460,"speech_dominance":0.3134,"duration_seconds":1183.346}
+{"key":"conv_10cf7f6ff081","language":"en","relationship":"friends","topics":["evolution","human origins","biology","dna and genetics","scientific debate"],"turns_per_minute":3.704,"turn_taking_gap_ms":640,"speech_dominance":0.3077,"duration_seconds":1376.882}
+{"key":"conv_e11edcca8521","language":"en","relationship":"friends","topics":["prescription medication","celebrity cosmetic surgery","perfume and fragrances","consumer culture","weather and flooding","online shopping livestreams"],"turns_per_minute":16.147,"turn_taking_gap_ms":420,"speech_dominance":0.4935,"duration_seconds":1575.568}
+{"key":"conv_04692967d40e","language":"en","relationship":"friends","topics":["formula one","motorsport","sports viewing","live events","car racing rules"],"turns_per_minute":3.9,"turn_taking_gap_ms":760,"speech_dominance":0.6972,"duration_seconds":1015.274667}
+{"key":"conv_5a798c657bb0","language":"en","relationship":"friends","topics":["family relationships","weather","introversion vs extroversion","social encounters","extended family history","health scares"],"turns_per_minute":9.332,"turn_taking_gap_ms":380,"speech_dominance":0.36,"duration_seconds":1273.058667}
+{"key":"conv_2e4aec4666d8","language":"en","relationship":"friends","topics":["football","sports history","athletes","rivalries","world cup"],"turns_per_minute":4.968,"turn_taking_gap_ms":860,"speech_dominance":0.6895,"duration_seconds":1002.448}
+{"key":"conv_aa4777220a1e","language":"en","relationship":"friends","topics":["horror movies","1980s nostalgia","first jobs","retail work","music videos","child actors"],"turns_per_minute":7.89,"turn_taking_gap_ms":360,"speech_dominance":0.4431,"duration_seconds":3604.53866666667}
+{"key":"conv_b98b64206aee","language":"en","relationship":"friends","topics":["books and reading","self-improvement","artificial intelligence","formula one","motogp","endurance racing"],"turns_per_minute":5.082,"turn_taking_gap_ms":560,"speech_dominance":0.6829,"duration_seconds":1015.368}
+{"key":"conv_31982da49c73","language":"en","relationship":"friends","topics":["online shopping","thrifting","fashion","hair color","estranged family","weight loss"],"turns_per_minute":16.296,"turn_taking_gap_ms":380,"speech_dominance":0.5575,"duration_seconds":2724.676333}
+{"key":"conv_14f897f75018","language":"en","relationship":"friends","topics":["home appliances","formula one","auto racing training","basketball","nba finals","american sports culture"],"turns_per_minute":4.108,"turn_taking_gap_ms":640,"speech_dominance":0.441,"duration_seconds":1402.070667}
+{"key":"conv_f5c0b02f01e2","language":"en","relationship":"friends","topics":["trivia quizzes","geography","world history","science facts","food and drink","travel plans"],"turns_per_minute":19.77,"turn_taking_gap_ms":320,"speech_dominance":0.4089,"duration_seconds":3705.672}

preview/metadata.jsonl

@@ -1,16 +1,16 @@
-{"file_name":"conv_bcff0c2a0be7.mp3","topics":"halloween, horror movies, childhood memories, religion and upbringing, relationships, holiday traditions","duration_min":36,"turns_per_minute":9.8,"speaker_a":"45-54 · female · US · native English","speaker_b":"35-44 · male · US · native English"}
-{"file_name":"conv_756ee444d825.mp3","topics":"music and concerts, relationships, communication habits, retail work, personal style, perfume and scent","duration_min":74,"turns_per_minute":9.2,"speaker_a":"45-54 · female · US · native English","speaker_b":"35-44 · male · US · native English"}
-{"file_name":"conv_d4005da6db98.mp3","topics":"daily routines, travel planning, visa and paperwork, martial arts training, soccer, world cup","duration_min":29,"turns_per_minute":7.2,"speaker_a":"16-24 · male · FR · native English","speaker_b":"16-24 · male · GB · native English"}
-{"file_name":"conv_31bedfc99b53.mp3","topics":"soccer, world cup, sports predictions, gym and fitness, immigration policy, sports organizations","duration_min":23,"turns_per_minute":4.5,"speaker_a":"16-24 · male · FR · native English","speaker_b":"16-24 · male · GB · native English"}
-{"file_name":"conv_cb37687ec915.mp3","topics":"food and recipes, weather, relocation, parenting adult children, pets and cats, grief and loss","duration_min":43,"turns_per_minute":14.9,"speaker_a":"55-64 · female · US · native English","speaker_b":"35-44 · female · US · native English"}
-{"file_name":"conv_0c917fc63b3b.mp3","topics":"career changes, software startups, space travel, aviation, flat earth theory, conspiracy theories","duration_min":20,"turns_per_minute":2.7,"speaker_a":"16-24 · male · GB · native English","speaker_b":"16-24 · male · FR · native English"}
-{"file_name":"conv_10cf7f6ff081.mp3","topics":"evolution, human origins, biology, dna and genetics, scientific debate","duration_min":23,"turns_per_minute":3.7,"speaker_a":"16-24 · male · GB · native English","speaker_b":"16-24 · male · FR · native English"}
-{"file_name":"conv_e11edcca8521.mp3","topics":"prescription medication, celebrity cosmetic surgery, perfume and fragrances, consumer culture, weather and flooding, online shopping livestreams","duration_min":26,"turns_per_minute":16.1,"speaker_a":"55-64 · female · US · native English","speaker_b":"35-44 · female · US · native English"}
-{"file_name":"conv_04692967d40e.mp3","topics":"formula one, motorsport, sports viewing, live events, car racing rules","duration_min":17,"turns_per_minute":3.9,"speaker_a":"16-24 · male · FR · native English","speaker_b":"16-24 · male · GB · native English"}
-{"file_name":"conv_5a798c657bb0.mp3","topics":"family relationships, weather, introversion vs extroversion, social encounters, extended family history, health scares","duration_min":21,"turns_per_minute":9.3,"speaker_a":"45-54 · female · US · native English","speaker_b":"35-44 · male · US · native English"}
-{"file_name":"conv_2e4aec4666d8.mp3","topics":"football, sports history, athletes, rivalries, world cup","duration_min":17,"turns_per_minute":5.0,"speaker_a":"16-24 · male · FR · native English","speaker_b":"16-24 · male · GB · native English"}
-{"file_name":"conv_aa4777220a1e.mp3","topics":"horror movies, 1980s nostalgia, first jobs, retail work, music videos, child actors","duration_min":60,"turns_per_minute":7.9,"speaker_a":"45-54 · female · US · native English","speaker_b":"35-44 · male · US · native English"}
-{"file_name":"conv_b98b64206aee.mp3","topics":"books and reading, self-improvement, artificial intelligence, formula one, motogp, endurance racing","duration_min":17,"turns_per_minute":5.1,"speaker_a":"16-24 · male · FR · native English","speaker_b":"16-24 · male · GB · native English"}
-{"file_name":"conv_31982da49c73.mp3","topics":"online shopping, thrifting, fashion, hair color, estranged family, weight loss","duration_min":45,"turns_per_minute":16.3,"speaker_a":"55-64 · female · US · native English","speaker_b":"35-44 · female · US · native English"}
-{"file_name":"conv_14f897f75018.mp3","topics":"home appliances, formula one, auto racing training, basketball, nba finals, american sports culture","duration_min":23,"turns_per_minute":4.1,"speaker_a":"16-24 · male · GB · native English","speaker_b":"16-24 · male · FR · native English"}
-{"file_name":"conv_f5c0b02f01e2.mp3","topics":"trivia quizzes, geography, world history, science facts, food and drink, travel plans","duration_min":62,"turns_per_minute":19.8,"speaker_a":"45-54 · male · US · native English","speaker_b":"55-64 · male · TH · native English"}
+{"file_name":"conv_bcff0c2a0be7.mp3","topics":"halloween, horror movies, childhood memories, religion and upbringing, relationships, holiday traditions","relationship":"friends","duration_min":36,"turns_per_minute":9.8,"turn_taking_gap_ms":400,"speech_dominance":0.4}
+{"file_name":"conv_756ee444d825.mp3","topics":"music and concerts, relationships, communication habits, retail work, personal style, perfume and scent","relationship":"friends","duration_min":74,"turns_per_minute":9.2,"turn_taking_gap_ms":340,"speech_dominance":0.4}
+{"file_name":"conv_d4005da6db98.mp3","topics":"daily routines, travel planning, visa and paperwork, martial arts training, soccer, world cup","relationship":"friends","duration_min":29,"turns_per_minute":7.2,"turn_taking_gap_ms":440,"speech_dominance":0.51}
+{"file_name":"conv_31bedfc99b53.mp3","topics":"soccer, world cup, sports predictions, gym and fitness, immigration policy, sports organizations","relationship":"friends","duration_min":23,"turns_per_minute":4.5,"turn_taking_gap_ms":640,"speech_dominance":0.45}
+{"file_name":"conv_cb37687ec915.mp3","topics":"food and recipes, weather, relocation, parenting adult children, pets and cats, grief and loss","relationship":"friends","duration_min":43,"turns_per_minute":14.9,"turn_taking_gap_ms":440,"speech_dominance":0.49}
+{"file_name":"conv_0c917fc63b3b.mp3","topics":"career changes, software startups, space travel, aviation, flat earth theory, conspiracy theories","relationship":"friends","duration_min":20,"turns_per_minute":2.7,"turn_taking_gap_ms":460,"speech_dominance":0.31}
+{"file_name":"conv_10cf7f6ff081.mp3","topics":"evolution, human origins, biology, dna and genetics, scientific debate","relationship":"friends","duration_min":23,"turns_per_minute":3.7,"turn_taking_gap_ms":640,"speech_dominance":0.31}
+{"file_name":"conv_e11edcca8521.mp3","topics":"prescription medication, celebrity cosmetic surgery, perfume and fragrances, consumer culture, weather and flooding, online shopping livestreams","relationship":"friends","duration_min":26,"turns_per_minute":16.1,"turn_taking_gap_ms":420,"speech_dominance":0.49}
+{"file_name":"conv_04692967d40e.mp3","topics":"formula one, motorsport, sports viewing, live events, car racing rules","relationship":"friends","duration_min":17,"turns_per_minute":3.9,"turn_taking_gap_ms":760,"speech_dominance":0.7}
+{"file_name":"conv_5a798c657bb0.mp3","topics":"family relationships, weather, introversion vs extroversion, social encounters, extended family history, health scares","relationship":"friends","duration_min":21,"turns_per_minute":9.3,"turn_taking_gap_ms":380,"speech_dominance":0.36}
+{"file_name":"conv_2e4aec4666d8.mp3","topics":"football, sports history, athletes, rivalries, world cup","relationship":"friends","duration_min":17,"turns_per_minute":5.0,"turn_taking_gap_ms":860,"speech_dominance":0.69}
+{"file_name":"conv_aa4777220a1e.mp3","topics":"horror movies, 1980s nostalgia, first jobs, retail work, music videos, child actors","relationship":"friends","duration_min":60,"turns_per_minute":7.9,"turn_taking_gap_ms":360,"speech_dominance":0.44}
+{"file_name":"conv_b98b64206aee.mp3","topics":"books and reading, self-improvement, artificial intelligence, formula one, motogp, endurance racing","relationship":"friends","duration_min":17,"turns_per_minute":5.1,"turn_taking_gap_ms":560,"speech_dominance":0.68}
+{"file_name":"conv_31982da49c73.mp3","topics":"online shopping, thrifting, fashion, hair color, estranged family, weight loss","relationship":"friends","duration_min":45,"turns_per_minute":16.3,"turn_taking_gap_ms":380,"speech_dominance":0.56}
+{"file_name":"conv_14f897f75018.mp3","topics":"home appliances, formula one, auto racing training, basketball, nba finals, american sports culture","relationship":"friends","duration_min":23,"turns_per_minute":4.1,"turn_taking_gap_ms":640,"speech_dominance":0.44}
+{"file_name":"conv_f5c0b02f01e2.mp3","topics":"trivia quizzes, geography, world history, science facts, food and drink, travel plans","relationship":"friends","duration_min":62,"turns_per_minute":19.8,"turn_taking_gap_ms":320,"speech_dominance":0.41}