Zoom Logo

Unlocking more from your Audio Data - Shared screen with speaker view
Jayson DeLancey (Dolby)
14:06
https://github.com/Briggs599/audio_data_DolbyIO
Braden Riggs
14:31
https://github.com/Briggs599/audio_data_DolbyIO
Eyal Kazin
15:41
Thanks for introducing 548 style!
Eyal Kazin
20:37
And for putting the adds that you mentioned …
Tom Niesytto
20:45
is that a common format for transcribed audio?
Eyal Kazin
21:44
Maybe they did freestyle rap …
Tom Niesytto
24:12
could you zoom it slightly?
Marcelo Condori. PyData Global Moderator
24:34
Hello Tom, would you specify the format your are referring in your question, please?
Tom Niesytto
25:48
can you show code
Olga Lyashevska
25:55
Code was a little small
Tom Niesytto
25:56
nice!
Jim Sheldon
27:31
What is the rough cost in time and money for Azure to transcribe an hour-long podcast?
Minas Sifakis
28:32
Sorry, I joined a bit late. Can someone please post the location of the notebooks?
Tom Niesytto
28:41
sarcasm is GONE.
Jayson DeLancey (Dolby)
28:44
Here’s the notebook on screen now if you want to see code: https://github.com/Briggs599/audio_data_DolbyIO/blob/main/analysis.ipynb
Marcelo Condori. PyData Global Moderator
28:45
https://github.com/Briggs599/audio_data_DolbyIO
Tom Niesytto
32:08
Could you simply extract unique identifiers that correspond to real life object and correlate with frequence of mentions on Twitter - then you would know if the episode was on popular subject
Jim Sheldon
35:01
Thank you!
Braden Riggs
36:30
https://github.com/tyiannak/pyAudioAnalysis
Tom Niesytto
42:52
It would make sense line it up with transcribed text so the energy is on per-sentence basis - you could analyze how specific words are stressed perhaps?
Eyal Kazin
47:39
Nice visual! What is the name of this type of plot?
Eyal Kazin
51:56
Found it! chord diagram
Eyal Kazin
56:48
Did it identify the different speakers, or was that part of the meta data?
Tom Niesytto
58:39
would be interesting to run it on comedian doing imporessions
Eyal Kazin
59:02
Tom, I like the way you think ;-)
Tom Niesytto
59:14
likewise ;-)
Eyal Kazin
01:01:13
Overlapping tracks ...
Tom Niesytto
01:02:05
I would be curious how those packages are tested
Eyal Kazin
01:05:02
Speaker cliques as a function of time is pretty cool to visualise !
Tom Niesytto
01:06:02
ferris bueller...
Tom Niesytto
01:07:14
train people to be better public speakers
Marcelo Condori. PyData Global Moderator
01:09:00
https://go.dolby.io/pydata-2021
Marcelo Condori. PyData Global Moderator
01:09:24
Braden's link
Braden Riggs
01:10:03
https://go.dolby.io/pydata-2021
Tom Niesytto
01:15:49
Just one project idea?
Eyal Kazin
01:25:40
Or sneezing!
Eyal Kazin
01:25:44
:-)
Tom Niesytto
01:26:07
can you do foreign languages?
Tom Niesytto
01:29:49
In transcribing - how is punctuation inserted?
Tom Niesytto
01:30:26
For example how does it recognize question?
Tom Niesytto
01:31:09
punctuation is inserted vs punctuation is inserted?
Tom Niesytto
01:33:18
GREAT talk - thank you **3
Olga Lyashevska
01:33:23
great talk, thanks!
Eyal Kazin
01:33:37
Thanks! Great material and delivery. Keep it up!
Jim Sheldon
01:33:52
πŸ‘
Ignacio Ruiz
01:34:08
great presentation! πŸ‘πŸ’‘
Jayson DeLancey (Dolby)
01:34:14
πŸ‘
Braden Riggs
01:35:02
https://github.com/Briggs599/audio_data_DolbyIO
Braden Riggs
01:35:30
https://github.com/DolbyIO/awesome-audio