dChan
Anonymous ID: 7d14c3 March 25, 2019, 11:38 p.m. No.5898458   🗄️.is 🔗kun   >>8772 >>8948 >>9048 >>9095

Reposting since late /lb:

 

Flowers v. Mississippi Closed-Captions

 

Here's sauce to the closed-captions from the c-span video Q linked:

 

https://www.c-span.org/fragments/convertCap.php?progid=521988

 

Format is ugly. Only advantage is its easy to line up with video time.

 

>>5898360 /pb

 

Another anon posted the sauce, which has a much tastier format:

 

https://ballotpedia.org/Flowers_v._Mississippi

 

>here is a transcript of the oral arguments. RBG speaks several times throughout, not just at the time Q pointed us to. first time is at 8:08. so there must be something particularly important about whatever happens at 19:56. maybe include in notables?

Anonymous ID: 7d14c3 March 25, 2019, 11:51 p.m. No.5898570   🗄️.is 🔗kun   >>8772 >>8948 >>9048 >>9095

>>5898464

>>5898474

>>5898487

 

All this RGB talk has got me thinking of the following vice article that made the notables a couple weeks back…

 

'''‘Deep Voice’ Software Can Clone Anyone's Voice With Just 3.7 Seconds of Audio

Using snippets of voices, Baidu's ‘Deep Voice’ can generate new speech, accents, and tones.'''

 

https://motherboard.vice.com/en_us/article/3k7mgn/baidu-deep-voice-software-can-clone-anyones-voice-with-just-37-seconds-of-audio

 

Anons have any ideas on how to approach detecting deep audio fake like this?

Anonymous ID: 7d14c3 March 26, 2019, 12:49 a.m. No.5898997   🗄️.is 🔗kun   >>9048 >>9095

Audio Analysis with Praat Speech Analyser

 

Disclaimer: not an audio expert, never done forensic audio analysis, no idea what I'm doing… yet. Just willing to hack at things and learn while perusing bewbs, sipping covfefe, and learning about history, kek.

 

Ok, with keks now out of the way, I think anons can use opensource Praat to do audio analysis on RGB's voice from Q's c-span link and past known samples. Pic related, shows formants (and some other things) from the first ~5s of anons rbgvoicecomparisons.mp4 (converted to mp3), see >>5898464

 

From what I've been reading, Sonogram and Formant analysis look like a good starting; and Praat looks like it can do both, and much moar. Just beginning to dig in, and will look into it more tmrw, but sharing what I'm perusing at the moment.

 

Here's some learning sauces:

 

http://www.fon.hum.uva.nl/praat/

https://www.soundonsound.com/techniques/introduction-forensic-audio

http://what-when-how.com/forensic-sciences/voice-analysis/

https://www.phon.ucl.ac.uk/resource/sfs/howto/formant.php

 

Haven't been able to find a copy just yet but, if an anon could source this book and post sauce, looks like it would be very helpful:

 

Principles of Forensic Audio Analysis by Robert C. Maher (2010)