---
title: "Normalize Audio Levels"
video_id: O2eyKOFvZfM
url: https://www.youtube.com/watch?v=O2eyKOFvZfM
workflow_page: https://workflows-of-ai.com/workflow/normalize-audio-levels
language: en
captions: generated
---

0:00 hey there Marcus here today we're going
0:03 to talk about how we can automate audio
0:05 levels and make them better well
0:09 actually let's
0:11 restart hey there Marcus here today
0:14 we're going to talk about how we can
0:16 automate audio levels and make them
0:20 better so I think you got the idea what
0:23 it is we're trying to do here by
0:25 listening to this but if you didn't
0:28 aonic is basically your your AI sound
0:31 engineer and I have created a module for
0:33 this so aonic is a service that takes
0:36 audio files and normalizes it it can do
0:39 a lot more but that's the integration
0:42 that is built into Drupal currently uh
0:45 and this is of course very important if
0:47 you for instance have many files with
0:49 different uh audio levels so this is
0:53 something I'm very bad at uh I'm using a
0:56 very hobby is microphone right so so
1:00 you might not get the best quality from
1:03 from when I'm talking hopefully you can
1:06 understand me at least uh so if you have
1:09 multiple files that has different audio
1:11 levels and you want to normalize and get
1:13 better quality this is a solution for
1:16 this it's also you know if you're
1:18 interviewing someone and you might only
1:20 have one microphone and you might point
1:21 it at that person's face then the
1:24 questions you're trying to ask might not
1:26 be the best quality this of course
1:29 happens a lot if you watch druple
1:31 presentations it might be that someone
1:33 from the audience is asking a question
1:35 and because the microphone is far away
1:37 you only hear this very faintly so this
1:40 is also something that you can try to
1:42 fix with a
1:43 phonic um and we're going to automize
1:47 this so phonic is uh a module I built
1:50 for the AI
1:51 interpolator uh that basically takes a
1:54 input file and uh generates output file
1:58 that's normalized and that's it so it's
2:00 very simple uh that's what we're going
2:03 to set up now before I start with that I
2:06 would just want to uh uh to remind
2:09 people about workflows of AI so if you
2:11 don't subscribe to the YouTube channel
2:14 you can go to workflows of AI and there
2:17 you can search and filter by different
2:19 modules where I show different workflows
2:21 AI workflows connected to
2:24 troo um if you don't know druple Dro is
2:27 open source free software all the
2:29 modules I show here are open source and
2:32 free of course they're using um services
2:36 that cost money so in this case it's the
2:38 phonic service however I will mention
2:41 this you actually get two hours uh free
2:44 every month which means that if you're
2:47 if you're only doing some simple
2:50 podcasting or something like that that's
2:52 not too long you can actually use
2:55 this the other thing uh that I want to
2:58 show as well
3:00 is uh uh this so this is another reason
3:05 why I did this or this is actually the
3:07 main reason so there is a completely
3:09 100% automated um podcast called the
3:13 Drupal Weekly News on podcast. workflow
3:16 of- a.com or uh in most most podcast
3:22 software you can just search for drup of
3:24 Weekly News Spotify not included but
3:27 almost everything else and the problem
3:29 with this it's using audio from 11 Labs
3:32 then uh that might come at different
3:35 levels and also when I uh concatenate
3:38 this using ffmpeg there might be uh
3:41 level differences which was pretty um
3:45 annoying uh so the idea now is to
3:48 implement this into the workflow of
3:51 podcasts uh Drupal Weekly News podcast
3:54 so there's uh normalized good quality so
3:57 it sounds as professional as a real
4:00 podcast in terms of Audio Level as
4:04 well so uh I'm just going to give
4:07 example how you use this it's actually
4:09 very very easy so uh we're just going to
4:13 start by
4:15 downloading uh uh the files we need need
4:19 in this case we only need two we need
4:21 the
4:22 aonic uh so I'm running DD so I have to
4:25 set ddev in front and then we just need
4:28 the AI interpolator module as well as
4:29 well and the reason for this is that the
4:31 phonic is just a core service module and
4:35 then it has the interpolated module and
4:37 the idea is there that maybe there comes
4:41 uh other Integrations that doesn't have
4:43 anything to do with uh the interor so I
4:47 wanted to make sure to split this then
4:50 we just log into our website after we
4:52 got these two modules click extend and
4:55 we're just going to search for the aonic
4:58 AI interpolator phic this will install
5:00 no phonic core it will install AI
5:04 interpolator uh that's the first step
5:06 then we just need to set it up so it's
5:08 under configuration web services ofon
5:11 config and I am just going to get my
5:15 username because I don't really know
5:18 what it is here uh so it's not your your
5:23 uh
5:24 mail it's uh another username so this is
5:29 my user username uh and then my password
5:33 so it doesn't work with API key it works
5:36 with username password right so after I
5:39 have set up this I'm just going to go to
5:41 content types and add a content type
5:43 called normalize
5:47 audio uh in this case it's just helping
5:51 a lot so I'm just a helper functions I'm
5:54 just going to uh turn off all the stuff
5:57 there right uh uh for a typical
6:01 publishing workflow I remove the body
6:03 field and I'm just going to create two
6:05 field file now so I'm going to start
6:07 with input file which is a file that
6:11 takes MP3s we're just going to upload
6:14 one in this case but in theory you could
6:15 upload multiple
6:17 and uh basically allow uh multiple
6:21 outputs then so we have an input file
6:24 and the thing we're going to uh the
6:27 thing we're going to automate is the
6:28 output file
6:30 so we just take uh file in this case uh
6:34 MP3 and then we enable the AI
6:36 interpolator so this is where we're
6:38 going to find our Rule and we have in
6:40 this case aonic normaliz audio and we're
6:44 run base mode and we connect it to the
6:46 input file so this is basically telling
6:48 it take the input file normalize that
6:50 and put it in the output
6:52 file if you're going to work with large
6:55 files this can take some time so make
6:57 sure you're using the Chron process work
6:59 then so you don't use the direct or or
7:02 batch because they unless you have a
7:04 long server time out they will basically
7:06 time out before they're finished but in
7:09 this case I'm just going to use the
7:11 intro file that I used before uh to
7:16 generate this so we're just going to add
7:18 our content now then normalize audio
7:21 test just going to add a file that's the
7:25 intro. MP3 I'm just going to play it
7:28 here so you hear the quality of
7:31 it hey there Marcus here today we're
7:34 gonna talk about how we can automate
7:36 audio levels and yeah you get the idea I
7:39 click
7:42 save okay I did not put in my password
7:47 correct let's try this
7:51 again
7:53 uh let's try another password if this is
7:57 the correct one we save
8:01 H now it seems to be working that seems
8:03 to work
8:06 better uh yeah so the problem with using
8:09 password is there is no API key that I
8:12 can click on that has um that has the
8:16 API key so I have to remember the
8:18 password uh so now we have output file
8:21 we'll listen to
8:22 this hey there Marcus here today we're
8:25 going to talk about how we can automate
8:28 audio levels and make them better well
8:31 actually that's so you here it took up
8:35 the volume a little bit um and yeah
8:40 that's basically it very very simple
8:43 workflow today um it's not more complex
8:46 than this this but of course if you work
8:48 a lot with audio video and you need to
8:51 reset uh the levels or you're working
8:53 with automated podcasts uh
8:57 automated um audio files for for blind
9:02 people for instance then you can get a
9:05 nice normalization using this service
9:08 and it's as I said you get two hours for
9:09 free so it doesn't hurt to test it even
9:12 if you might not want to pay uh for it
9:15 so that's it thank you for watching