How to Record Audio and Graph wave .wav files using Python!

what's up you guys welcome back to the channel in today's video i'm going to show you how to record audio and graph wave files using python [Music] so i've been traveling a lot for work lately and i'm not at home this week so i don't have my multiple screens and setups to do like a real detailed line-by-line tutorial but i wanted to bring you guys this cool quick project that uses recording audio and the matplotlib graphing library to do some really cool audio visual stuff so i'm going to do a slightly different format for this video where i'm going to show the code that i already have written and explain what each section does but i'm not actually going to type it out line by line in real time with you guys let me know in the comments below if you like this format more or if you find the tutorials where we do go line by line and i explain it as i write it let me know what your preference is and i'll start doing more of that type of tutorial if people lean one way or another otherwise i'll probably just keep a good mix on the channel so thank you guys so much for your feedback and without any further ado let's start talking about the project so to start you're going to want to import the pi audio module and you should be able to do this just by using dash m pip install pi audio there's a lot of stack overflow and online help about how to install these modules so if you're having any trouble with any of the ones you see here just check out the internet i'm not going to go too in-depth about how to run pip because hopefully by this stage of project building you know how to import modules but you need the pi audio module to actually record from your computer and then you want to import the wave module which most of the time is a built-in module that comes installed with your python install and that's the python module that lets you parse wav files and process them and use them which we'll need to do in this project then import the matplotlib.pyplot as plot i always alias it you don't technically have to do this but trust me it saves you a lot of typing later on this is the graph package that's actually going to give us the visual of our wave file when we record and then import numpy as np honestly even if you weren't doing this project in particular this is a package i import at the beginning of almost every project because it's got so many shortcuts for math and arrays and data visualization it's a great one to have in your projects and then this second block of four rows is us setting up some specifics about our audio files and i will touch on what each of them do real quickly but just know you can kind of use these as a default if you don't really understand what we're looking at frames per buffer is essentially how many frames are going to be recorded per second and you'll notice how we use that when we get down to actually recording um but 3200 is super standard and we need to put it kind of in that format because when we get to the point where we open the computers like stream and record from the microphone or your computer microphone whatever you're using that's the format it uses if you're peaking ahead a little bit here you'll see there are some specific parameters that we give the stream when we open pi audios like microphone recording input and so just use frames per buffer as 3200 for now if you don't want to dive too deep into why that is and then the format this is where these next two are kind of together if you're recording in what's called mono or single channel recording then you want your format to be that of p816 if you were recording in stereo which would have two channels you would want the pa in 32 but we're just going to record in mono because every microphone can handle that not every microphone can handle recording in stereo and then for rate this is going to be 1600 that is a really standard wave recording frequency um so that's what i'm going to use you don't have to but again i don't want to get too deep into the actual science of audio files i just want to show you guys how to write the code so then i just like to create this alias variable where i say pi audio dot pi audio with parentheses is equal to pa because typing this every time you want to access something from pi audio is a bit messy and this makes it a lot cleaner and neater to read your code but so then this this stream is actually where we open the computer and access the microphone essentially so from pi audio it's just pa.open and then those variables we set up before format channels rate and frames per buffer this looks kind of like kivy if you've designed any graphical user interfaces in qivi there's a lot of parameters that you just set you know parameter name equal to and if you're setting up a big project it's smart to just use a variable in there so that you can play around with the variables up here instead of having to mess around inside your stream so anyways just format channels rate and frames per buffer and then anytime you're trying to access the microphone set input equal to true and then uh because i want kind of a countdown on the screen while you're recording we're going to just print the word start recording on screen and now anything you say after that gets printed is in our kind of our recording so uh then we set up some new variables seconds is gonna be how long you want the recording so for me i just picked eight seconds because you can say full sentences in there but it doesn't go on too long this would be a tedious video if you guys had just had to watch me record minutes of data but so if you want longer recordings you can make this a larger value create an empty list for your frames and frames is actually going to be individually capturing the audio frequency of your recording device every single iteration so when we set our rate divided by our frames per buffer that's going to be how many times per second it's recording so that's quite a large number um and so you're going to store that all in a list that we're going to use both for storing the wav file later and for graphing it at the end and then these two second tracking and second count these are definitely optional but i like the idea of in the terminal just showing how much time is left in your project um in your recording rather and so what i do is i create this variable second tracking that's going to add one every single time we record a new frame but then once we've recorded enough frames that equal one second so remember i said that's the rate at which we're sampling divided by the number of frames per buffer that's going to tell you how many seconds you have or that's going to tell you how many frames you have in one second so if second tracking is equal to that value that means we've gone one second i reset my second tracking code and then i just add one to my counter for seconds and then i show how much time is left by printing the total seconds minus the number of seconds that have passed this is just a nice thing so that while you are recording um it's showing you you have seven seconds left six seconds left five seconds left if you want to it's not even a bad idea to say seconds afterwards so you don't have to guess about units but then once it finishes so this means we've gotten to the point where we've done this many which is how many frames per second times the total number of seconds and once we get to that point you do stream dot stop stream stream dot close and then pi audio which we've aliased as pa dot terminate and this is everything you need to shut down the recording but then what we want to do is we want to store the information in a file so what i do is i create another alias variable that i call object or obj and then set it equal to wave dot open so again the wave module is like the built-in package for parsing data and then give it a name so this actually doesn't have to be an existing audio file if you see right now i do have that but if i delete this out because i just did a recording so i'm gonna delete it and i'm not gonna do the safe delete so i just got rid of it this code is still gonna run fine because we're not doing a read function right now we're doing a write function so the two things you need to give this wave.open argument are the name you want to give your wav file and then if you're writing this is wb for write binary if you were reading from it it would be rb which is read binary and if you're peaking ahead we actually do that later once we read it into graph it so i do it that way so that i can show you using the wave module a couple times but technically you do already have the data in this file so then what you do is with the wav file you're already creating a wave object by saying wave.open so there are these built-in functions that you need to populate with those things we've already created so set the number of channels and this is what's going to tell the wave file if it's a mono or stereo and then set sample width is just telling it the total size and by doing that it just needs to know the format so it's able to figure out how many frames are in it by knowing what format so if this is a stereo recording that's going to be in 32 but if it's a mono recording it's in 16 and all you have to do is just set the sample width by doing pi audio dot get sample size of the format and then object dot set frame rate and then just tell it the frame rate that you were recording at and then the last thing you do is you say object dot write frames and then you do b uh parenthesis parenthesis dot join and then all of the frames and so this is the list we just recorded so this is our full list of all of the data while we were recording and we just joined that to this right frames object module that's not the framework function of the wave module and then just close it so this object.close is what actually saves all of your work and now you're going to have a file so if you were to stop the program right at this block of code you're going to have an audio file over here everything after that code is now just so that we can graph it on the screen if you were to stop right there you would have everything you needed to get a recording so then what i do now to kind of show you the flip side of things if you already had a recording and you wanted to graph it then start with file equals wave.open and the only reason i say file instead of object again is to kind of distinguish the two functions but you could call this object again if you needed to or you don't really have to open it back up since it already exists but again i'm trying to separate out the two functionalities so we're going to open the file right after we record it and notice now it's rb because it's read binary now we're just getting the data out of it and set up a few things so sample frequency is going to be file.getframe frames is going to be get number of frames and then the the actual wave is going to be reading the frames and then just give it a minus one which basically means read everything and then we're going to close it because now we've moved all the information we need into a into our local variables so we don't have to mess with the file anymore we have everything we need right here the total time of the recording is equal to frames divided by the sample frequency so that tells us how long it is and then again if one channel use int 16 here for data type otherwise if it's stereo use int 32 but we're creating an array and here's why numpy is great for this project um we're just going to create an array and then we say from a buffer and then we give it the wave and we say what data type it was produced with um and so that's all we have to do as long as you follow the same steps that i gave you here that's going to give you an array of all of the data that was recorded and it knows how to graph that so then the times are going to be equal to numpy again and now this is lin space l in space um and then zero up to time where the number is equal to frames so this is saying your start time and your end time and then the total number of data points to expect in that space and so this should be logical because time is equal to the frames divided by the sample frequency which basically means from zero to time the frames are how many data points we have so the rest of this is just creating the graph so we use plot which again is from map plot lib dot pi plot so if you didn't do an alias in the beginning you're gonna need to type that out every time but assuming you're following along it would be plot dot figure and then the figure size and i just use 15 by five because it seems about right for an eight second recording you can play around with this this is just setting the parameters for the size of your screen and then you tell it what to put on the x-axis and then what the actual data is for the y-axis and so for us we want time in the x-axis and then we want the audio array in the y axis i just give it some labels so y label and x label for the signal wave and then the times and then the x limit is from zero to time and then give it a title and do plot.show and that's everything so now if we record a new one we wait till we get start recording there we go and now we're recording so i'll take a pause and now we'll talk a little bit more and you can see right here you know this doesn't really look like anything in particular because i was just talking over the video but you can see when i said let's take a pause and then we pretty much flat line and then we get to the end here and i'm talking again so you can do some kind of fun things with this and you can sort of see let's do that again so if i take big spaces will see some fun gaps between the words so you can see that last one it was much closer because i was sort of talking a normal pace but here i took big spaces between every word and you can see you get this kind of fun shape so you don't have to follow along with the video any longer and watch me just record multiple things if you wanted to um always record with a new uh a new file name then you could potentially set this up to be a little more robust to like running multiple times and handling different data every time for me this is plenty um basically what this is going to do though every time is overwrite the previous wave file so if you were gonna do multiple recordings and you want it with a new name you either gonna have to come in here and punch something manually or just create like a a start prompt where you get the name of the wav file as an input so um if you wanted to you would just do like up here input you know and then get uh enter file name something like that and then instead of you would call this like new name and instead of uh just writing to le master tech every time you would write to new name dot wave and you would have to do that as a formatted string so you would do something like that i'm not doing it that way i did just i guess show you how to do the whole thing so hopefully if you would prefer to do it that way go ahead i will probably throw all this code in a github and leave a link to it in the description below so if you don't feel like typing out everything you saw in here that's totally fine again let me know if you liked this kind of format where it was just an explanation of already written code or if you'd like to see the code written line by line with me so i was certainly happy to take feedback and make the channel better and with that being said if you have any ideas for future videos or things you'd like to see on the channel be sure to let me know in the comments below because i've been getting a ton of great suggestions and i am working on a lot of those projects so i really appreciate the support if you've enjoyed the video please consider leaving a like on the channel subscribe to the channel we're just getting started and we make a lot of cool projects so thank you everybody for the support good luck with your projects and thanks for watching thanks bye