Episode 11 #72

sockeqwe · 2017-05-15T14:47:30Z

Feed is valid: http://castfeedvalidator.com/?url=https://raw.githubusercontent.com/artem-zinnatullin/TheContext-Podcast/episode11/feed.rss

mp3 file url must be: https://artemzin.com/static/thecontext/episodes/The.Context.episode.11.mp3

TODO:

Add mp3 file to release section
Add git tag after release

artem-zinnatullin · 2017-05-15T15:59:38Z

@ming13 please review

artem-zinnatullin

Few small additions

artem-zinnatullin · 2017-05-15T16:02:52Z

how-to-record-and-publish.md

-11. Validate resulting RSS feed, for example w/ [this tool](http://castfeedvalidator.com)
+10. Update README: add new Episode information
+11. Post a PR with show notes and RSS feed update.
+12. Validate resulting RSS feed, for example w/ [this tool](http://castfeedvalidator.com)


artem-zinnatullin · 2017-05-15T16:03:35Z

show_notes/Episode_11.md

+  - [RxJava 2: Discussion about null values](https://github.com/ReactiveX/RxJava/issues/4644#issuecomment-256684743)
+  - [Reactive Streams specification: null value](https://github.com/reactive-streams/reactive-streams-jvm/issues/204)
+  - [Why RxJava 3: Split the library into two or adding types for symmetry](https://github.com/ReactiveX/RxJava/issues/4564)
+  - [RxJava 3 Preview](https://github.com/akarnokd/RxJava3-preview)


[RxJava 2 Kotlin SAM issue combineLatest()/etc](https://github.com/ReactiveX/RxKotlin/issues/103)

artem-zinnatullin · 2017-05-15T16:08:36Z

show_notes/Episode_11.md

+  - Artem Zinnatullin [@artem_zin](https://twitter.com/artem_zin), [personal blog](http://artemzin.com), [GitHub](https://github.com/artem-zinnatullin)
+
+#### Additional links:
+


[Juno's Android Rider Team RxJava 2 Migration Specification](https://github.com/gojuno/engineering/blob/master/specs/rxjava-2-migration.md)

It should be there at the time of publishing the episode, for now it's in the PR gojuno/engineering#1

arturdryomov

Looks great!

arturdryomov · 2017-05-15T17:06:49Z

feed.rss

+			<li>Flowable or Observable?</li>
+			<li>Other Rx Types: Single, Maybe and Completable</li>
+			<li>RxJava 2 interop library</li>
+			<li>RxJava 2: Default UncaughtExceptionHandler logs exceptions silently in jvm unit tests</li>


Nitpick — JVM

arturdryomov · 2017-05-15T17:07:38Z

show_notes/Episode_11.md

+ - Flowable or Observable?
+ - Other Rx Types: Single, Maybe and Completable
+ - RxJava 2 interop library
+ - RxJava 2: Default UncaughtExceptionHandler logs exceptions silently in jvm unit tests


Nitpick — JVM

good catch. Changed.

artem-zinnatullin · 2017-05-16T07:25:30Z

show_notes/Episode_11.md

+
+#### Guest:
+
+ - Artur Dryomov [GitHub](https://github.com/ming13)


We need your twitter handle, @ming13, it's a blocker!

sockeqwe · 2017-05-16T07:50:19Z

@artem-zinnatullin I have listened to latest Fragmented Podcast episode to compare the volume level with our episode. Fragmented is a little bit louder. Probably we should increase the volume of the whole file by 3-4 db. Unfortunately, I haven't audition on my machine at work. So if you do have audition installed and have 2 minutes of free time because of gradle build 😄 it would be great if you could increase the volume of the whole mp3 file by 3-4 db. Otherwise, I will do it this evening at home.

johnjohndoe · 2017-05-16T08:01:15Z

The major issue with the sound is not the overall sound level. It just becomes uncomfortable to listen to the show when the sound level of different speakers is different a lot - which is natural just by the fact how far you are away from the microphone, how loud you speak or which microphone you use. This becomes tedious when you listening to podcast on the street, in the train, ... what many people, that I know, do.
To overcome the issue you need to level out the average sound level of different speakers. This is more easy if you do multi-track recording. If you record all speakers as a single track then it is still possible. I just know Auphonic does a good job - but I am sure there is other services/software. You can listen to what the do here:

mur.strom podcast - MSP009 auphonic (DE)

No, I do not work for Auphonic :) I just learned about them in this podcast episode.

sockeqwe · 2017-05-16T08:23:37Z

Thanks @johnjohndoe for your feedback.
Well we tried this for our last episode: Here is our manually edited file:
https://www.dropbox.com/s/u3h1g1d3zrco23d/The.Context.episode.11.mp3?dl=0

And here is the automatically edited file by Auphonic:
https://www.dropbox.com/s/3af99goyykpt3gs/The.Context.episode.11-2-auto-edited.mp3?dl=0

Auphonic applied the average volume level, but the problem is that at the end of the episode our guest @ming13 was more away from the microphone then at the beginning and that some words of sentences were recorded poorly. With automatically avg. volume adjustment like Auphonic did it sounds much worse from my point of view.
So i think we will publish the manually edited episode where sound level might not be balanced through the whole episode, but from my point of view sounds better on those parts were recording quality was just very bad.

If you have time to listen at 40:00 - 50:00 of both files we will take your feedback into account.

arturdryomov · 2017-05-16T08:26:51Z

Sorry for the microphone issue guys. Really was trying to be steady and at the same place ;-(

sockeqwe · 2017-05-16T08:38:33Z

No problem: 95% of the audio is fine. Also quality of content is more important then audio volume imho. I just wanted to make a point that we might have to make some compromises while editing which may mean lower audio volume (or not averaged volume level) since we were not recording this in a perfect environment like a studio nor we have the knowledge and time to edit each second of the episode perfectly. However, if such automated tools exist like Auphonic and do a better job than we do, then of course we would just use them. However, imho Auphonic didn't work that well if the recorded files are already not perfect (which is almost always true for our podcast since we can't expect our guests to have professional microphones). Nevertheless, we are thankful for every feedback and suggestions. If someone knows a strong AI for audio editing, please let us know :) Artur Dryomov <notifications@github.com> schrieb am Di., 16. Mai 2017, 10:26:

…

Sorry for the microphone issue guys. Really was trying to be steady and at the same place ;-( — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#72 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAjnrvxhYW1t3g5wneSZc0-UpOJleGNzks5r6V3MgaJpZM4NbPZX> .

johnjohndoe · 2017-05-16T08:57:30Z

I listened to and compared both files. I definitely prefer The.Context.episode.11-2-auto-edited.mp3. The equal sound level makes it very pleasent to listen to it. To me there no decrease of the quality.

Also quality of content is more important then audio volume imho.

I disagree with you on this point. I imagine you loose a lot of listeners. When it is inconvenient to listen to the recording it is regardless how interesting the content might be.
To mention it - the Android Developers Backstage podcast had the same sound level problems. Their sound levels were so diverging that it was impossible to listen to their early episodes on the street with background noise.

Just to get more opinions: You might want to directly ask your listeners in the next show by running a survey or similar.

arturdryomov · 2017-05-16T08:58:41Z

ATP is my favourite podcast and there is a brief (and probably mostly outdated) explanation on how the podcast is made. Obviously Marco knows a lot about podcasts, he even created an app, so all editing mumbo-jumbo is on his side. Cannot find his story on the editing though.

sockeqwe · 2017-05-16T10:42:10Z

Thanks for your feedback @johnjohndoe .

Then @artem-zinnatullin we should ship this file:
https://www.dropbox.com/s/3af99goyykpt3gs/The.Context.episode.11-2-auto-edited.mp3?dl=0
I have relistened to the auto edited and indeed it is better. Yesterday I have heard some glitches and hard cuts on the auto edited file which I found disturbing, but I haven't noticed them anymore today. Maybe I was to tired yesterday.

@artem-zinnatullin don't forget to rename it to The.Context.episode.11.mp3 as this is the name / url in rss feed.

@ming13 the main difference is hardware I guess. Actually I think hardware matters the most. For example, my microphone is a semi professional (120$), Artem's one is about 65 $. For example in Episode 8 Damn functional programming Paco used a headset with microphone, mabe 30 $. We can't expect guests to have a semi professional microphone

That was one reason why in Episode 10: Kotlin Language Design Nitpicking, Dmitry Jemerov used my microphone to record his voice and my voice was a little bit less clear / volume was a little bit low.

The equipment of ATP podcast equipment:

150 $ Microphone
45 $ Microphone on/off switch
850 $ Sound device Audio interface for computer
10 $ Windsceen

So more than 1000$ in hardware and that is the major difference. Also they have a desk and fixed place to record the episode so that the distance between mic and his mouth is almost the same through the whole episode (which was the issue during this recording):

I think the major impact on sound quality is hardware. However, editing could improve the quality too. Since this is a very time consuming progress, we will try out a mix of both automated audio processing tools like Auphonic and manually audio editing for the next episodes. If it turns out that automated audio processing works quite good, I would be more than happy to spent some bucks on such services.

Thanks for your feedback and help!

artem-zinnatullin · 2017-05-16T11:43:34Z

@sockeqwe I have to agree with @johnjohndoe point and Android Developers Backstage example, lemme mix both of your opinions:

Sound quality should be good enough, however of course content quality is what we are aiming for.

I also like Auphonic version better, pricing of the service seems okay, @sockeqwe let's try to publish this version and ask users about results so we could decide whether we should process new episodes with this service or not?

artem-zinnatullin

Ok, green light from me, though @ming13 still does not want to sign up in twitter. @sockeqwe release when you want to, file is uploaded!

artem-zinnatullin · 2017-05-16T12:22:14Z

@sockeqwe thank you for hard work on releasing this episode! I hope from now we'll just need to combine tracks in one and process it through Auphonic and it'll take much less time!

sockeqwe · 2017-05-16T12:23:33Z

feed.rss

@@ -779,7 +779,7 @@
 		<atom:link rel="replies" type="text/html" href="https://github.com/artem-zinnatullin/TheContext-Podcast/issues/71"/>
 		<itunes:duration>52:46</itunes:duration>
 		<pubDate>Mon, 15 May 2017 09:30:00 EST</pubDate>
-		<enclosure url="https://artemzin.com/static/thecontext/episodes/The.Context.episode.11.mp3" length="50684128" type="audio/mpeg"/>
+		<enclosure url="https://artemzin.com/static/thecontext/episodes/The.Context.episode.11.mp3" length="50658006" type="audio/mpeg"/>


Exceptional good catch! 👍

I was very surprised that auto editing almost did not change file size, mp3 indeed is a very interesting format

because I used the already manually pre edited files for this episode. Auto editing results weren't that good with raw recorded mp3 files (much background noises and breathing after auto editing). So I think Auphonic was mostly adjusting volume and might have improved some pieces here and there.

sockeqwe · 2017-05-16T12:51:33Z

Well, not sure about it ... Tried Auphonic with other tracks episode ... Audition did a much better job on removing background noises and dynamic processing for removing echo and makes recorded voice clearer and bolder at the same time. Also our multitracks have different length depending on the record.
But this things are a 10 minutes manual editing job. So I think it could work like this in the future (at least I will try it for next episode):

MANUALLY: Use Audition to (should take 10 - 15 minutes)

remove background noises
remove echo / process dynamic
Align all recorded tracks (incl. cutting out recorded things before start and end)
Add intro and outro
remove pauses (300 ms). Removed over 8 minutes in total on this episode.
Export that as a single mp3 file.

AUTOMATICALLY: Let Auphonic process the exported file (result of manually step).
- Mainly adjusts the (average) volume
- Maybe / hopefully restores some poorly recorded parts

However, somethings (and those are the time consuming things) are still to be done manually (maybe after automatically processing):

Remove clicking / background noises, BREATHING is so annoying but hard to remove at the same time (time consuming).
Remove some things
Maybe remove some parts where two people are talking at the same time.

To be honest, I would like to not do that time consuming things at all (especially removing breathing) because we are not a professional podcast nor we earn any money in comparison for the time we put into producing the podcast. But this (time consuming) things are the important ones that make a good sound quality. It is mainly a hardware (microphone) issue of our guests. Auphonic didn't remove any of these things (breathing, clicking etc. at least with my test files) automatically. And those are the time consuming things while editing / processing a podcast.

Imho, we can't get rid of manually editing unless everybody has some good hardware (microphone).
Again, the simple tasks (listed in Manually 1) are not very time consuming, removing breathing , clicks, "ehm" (like in episode 9 with Alexey) are the time consuming things. I haven't seen automation for the later (time consuming) step.
Auphonic is (imho) just a last step to give it the final / finishing touches (at which it did a great job on this Episode 11) .

Btw. Autphonic suggests to export the mp3 file with 112 kBit/s which reduces the size of mp3 file for a few MB #70

Thanks everybody for your feedback and help. Really appreciated it.

artem-zinnatullin · 2017-05-16T20:51:01Z

@sockeqwe plan SGTM!

Episode 11

37963a1

sockeqwe added this to the Episode 11 milestone May 15, 2017

sockeqwe requested a review from artem-zinnatullin May 15, 2017 14:47

artem-zinnatullin requested changes May 15, 2017

View reviewed changes

sockeqwe added 2 commits May 15, 2017 18:51

Added Juno link and SAM issue link

904e011

Added git tag and upload to github release section

79a6d0b

arturdryomov approved these changes May 15, 2017

View reviewed changes

JVM - upper case

d4713ba

artem-zinnatullin reviewed May 16, 2017

View reviewed changes

Update "length" of the file in rss feed.

e970fc6

artem-zinnatullin approved these changes May 16, 2017

View reviewed changes

sockeqwe merged commit 2eb2dab into master May 16, 2017

sockeqwe deleted the episode11 branch May 16, 2017 12:17

sockeqwe commented May 16, 2017

View reviewed changes

sockeqwe mentioned this pull request May 17, 2017

Discussion Episode 11: Migration to RxJava 2 with Artur Dryomov from Juno #71

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Episode 11 #72

Episode 11 #72

sockeqwe commented May 15, 2017 •

edited

Loading

artem-zinnatullin commented May 15, 2017

artem-zinnatullin left a comment

artem-zinnatullin May 15, 2017

artem-zinnatullin May 15, 2017

sockeqwe May 15, 2017

artem-zinnatullin May 15, 2017

sockeqwe May 15, 2017

arturdryomov left a comment

arturdryomov May 15, 2017

arturdryomov May 15, 2017

sockeqwe May 15, 2017

artem-zinnatullin May 16, 2017

arturdryomov May 16, 2017

artem-zinnatullin May 16, 2017

sockeqwe commented May 16, 2017 •

edited

Loading

johnjohndoe commented May 16, 2017

sockeqwe commented May 16, 2017

arturdryomov commented May 16, 2017

sockeqwe commented May 16, 2017 via email

johnjohndoe commented May 16, 2017

arturdryomov commented May 16, 2017

sockeqwe commented May 16, 2017 •

edited

Loading

artem-zinnatullin commented May 16, 2017

artem-zinnatullin left a comment

artem-zinnatullin commented May 16, 2017

sockeqwe May 16, 2017

artem-zinnatullin May 16, 2017

sockeqwe May 16, 2017

sockeqwe commented May 16, 2017

artem-zinnatullin commented May 16, 2017

		- Artem Zinnatullin [@artem_zin](https://twitter.com/artem_zin), [personal blog](http://artemzin.com), [GitHub](https://github.com/artem-zinnatullin)

		#### Additional links:


		#### Guest:

		- Artur Dryomov [GitHub](https://github.com/ming13)

Episode 11 #72

Episode 11 #72

Conversation

sockeqwe commented May 15, 2017 • edited Loading

artem-zinnatullin commented May 15, 2017

artem-zinnatullin left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

arturdryomov left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sockeqwe commented May 16, 2017 • edited Loading

johnjohndoe commented May 16, 2017

sockeqwe commented May 16, 2017

arturdryomov commented May 16, 2017

sockeqwe commented May 16, 2017 via email

johnjohndoe commented May 16, 2017

arturdryomov commented May 16, 2017

sockeqwe commented May 16, 2017 • edited Loading

artem-zinnatullin commented May 16, 2017

artem-zinnatullin left a comment

Choose a reason for hiding this comment

artem-zinnatullin commented May 16, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sockeqwe commented May 16, 2017

artem-zinnatullin commented May 16, 2017

sockeqwe commented May 15, 2017 •

edited

Loading

sockeqwe commented May 16, 2017 •

edited

Loading

sockeqwe commented May 16, 2017 •

edited

Loading