mirror of
https://github.com/mollyim/webrtc.git
synced 2025-05-20 09:07:52 +01:00
77b376c094
Publicly available dataset of conversational speech audio recordings. This CL includes the following: - README.md: dataset description file, it also includes the scripts - *.wav.sha1: hash files for each audio track in the dataset The overall size of the wav files is ~36MB. The primary intended use of this dataset is in combination with the conversational speech tool (see https://chromium.googlesource.com/external/webrtc/+/master/webrtc/modules/audio_processing/test/conversational_speech/), using which longer recordings with custom turn switch timing can be created. BUG=webrtc:7218 Review-Url: https://codereview.webrtc.org/2869833002 Cr-Commit-Position: refs/heads/master@{#18068}
63 lines
2.9 KiB
Markdown
63 lines
2.9 KiB
Markdown
# WebRTC Conversational Speech dataset
|
||
|
||
This dataset includes *conversational speech* and *breathing* recordings,
|
||
and the audio files are named
|
||
``<langCode>_script<num>_<speakerGender>_sp<speakerId>_<turnId>.wav`` and
|
||
``breathing_<speakerGender>_sp<speakerId>_<num>.wav`` respectively.
|
||
|
||
The conversational speech part consists of *turns*. Each turn belongs to a
|
||
*script*, and a script involves two *speakers* named A and B.
|
||
|
||
The scripts have been designed spanning short and long turns. Also, the scripts
|
||
are gender neutral, which allows to let A and B turns be recorded by both male
|
||
and female speakers.
|
||
|
||
All the audio files have a single channel and a sample rate of 48000 Hz with 16
|
||
bit samples.
|
||
|
||
## English scripts (EN)
|
||
### Script 1
|
||
- A1: How is your day going?
|
||
- B1: Quite busy. I'm preparing for my presentation on our marketing strategy.
|
||
- A2: You must feel stressed out now.
|
||
- B2: That's an understatement.
|
||
### Script 2
|
||
- A1: Which book are you reading?
|
||
- B1: Let’s see if you can guess it, listen carefully. She tries to open the
|
||
doors. They are all locked; she begins to weep. She walks weeping to a high
|
||
glass table and sits down on its lower ledge. She sits on a big golden key and
|
||
picks it up in surprise. She tries it on all the doors but it does not fit.
|
||
- A2: Too hard, yet another hint.
|
||
- B2: Ok. “Would you tell me, please, which way I ought to go from here?” “That
|
||
depends a good deal on where you want to get to”, said the Cat. “I don't much
|
||
care where” she replied.
|
||
- A3: Stop, I know the rest! “Then it doesn't matter which way you go”, said the
|
||
Cat. “So long as I get somewhere” she added as an explanation. “Oh, you're sure
|
||
to do that, if you only walk long enough.”
|
||
- B3: So, what’s the title?
|
||
- A4: Alice in Wonderland!
|
||
- B4: Well done
|
||
## Swedish scripts (SE)
|
||
### Script 1
|
||
- A1: Hur är det?
|
||
- B1: Rätt upptagen. Jag förbereder mig inför min presentation om
|
||
marknadsföringsstrategi.
|
||
- A2: Då är du säkert riktigt stressad nu.
|
||
- B2: Det skulle man kunna säga.
|
||
### Script 2
|
||
- A1: Vilken bok läser du?
|
||
- B1: Jag ska läsa högt för dig. Försök gissa nu! “Dörren är ändå alldeles för
|
||
liten för mig. Och jag kan ju inte flyga upp igenom kaninhålet. Hon får syn på
|
||
en flaska på bordet. Drick mig. Bäst att vara försiktig. Om man dricker en
|
||
flaska där det står gift så får man nog ångra det förr eller senare. Men det
|
||
står ju bara drick mig.”
|
||
- A2: Det är för svårt! Får jag försöka igen?
|
||
- B2: Ok. “Åt vilket håll ska jag gå?” “Det beror på vart du vill komma”, sa
|
||
katten. “Det spelar ingen roll.” svarade hon.
|
||
- A3: Sluta, jag kan fortsättningen: “Då spelar det heller ingen roll åt vilket
|
||
håll du går." sa katten.
|
||
“…så länge som jag kommer någonstans”, fortsatte hon. “Åh, det kommer du att
|
||
göra”, sa katten, “bara du går tillräckligt långt”.
|
||
- B3: Så vad heter boken?
|
||
- A4: Alice i Underlandet
|
||
- B4: Jättebra
|