Month: November 2023

In order to summit my project, I have to pick and put the recording to gather. When I listen my recording, I feel I am listening through plants. It relies on the soil, like what plant do and temperature shape the audio as well. In order to show how the kit sound like, I don’t want to edit too much of my recoding. At first, I was thinking to simply cut some of the receding I have with his kit can put all together with gap between every audio to indicate the differences of those recoding. But soon, I found even I crossfade those track, the uniqueness of every track are so obvious, even two track are recorded in the same place.

One of the problem of this kit is accessibility. Even I soldier all the pot, button and resister, it still got a lot of jumper wires, and it have to work with a recorder as preamp as well. So every time when I start recording, I took me a long time to set up, have to connect mic to recorder, then connect output to Bela and connect all the jumper wires to Bela, and have to setup the shout cable moisture sensor and peg it into ground. So it would be better if I can build a boxes to put all together and build a preamp in the kit as well so doesn’t have to use it with recorder.

I went to few places to test out the Bela recoding kit, one of them is St Mary’s Churchyard (a park near LCC). I used my stereo pair mic Earsight Standard made by Immersive Soundscape and my Zoom F3. The field recorder is used as mic amplifier and the send to Bela thought line output. The recoding experience is fascinating. The sensation is weird where I was in the space that I very used to its soundscape: the busy traffic, the infrastructure, the train, etc. But listen through Bela, I feel like I was in different space. I was there but not really. The sound is familiar yet alienated. Even in the same small park, the spot I choose to peg my sensor in are completely different. Of course, it can be because of the position of my mics changed, but also there is more things to change the sound I am listening to.

To contrast different soundscape of outdoor space, I went back to LCC to find some plants to record with as it have to rely on the soil to make it works. The bridge in the library have one of the most interesting sound. The reverberation in there already huge, and with the reverb that added in, it create a weird sonic space. The indoor space also have different temperature, hence, the texture of the receding are completely different.

This is the final patch for this Bela Recorder patch. It receive stereo audio input from adc~ 1 and 2 and adc~ 3 content to *~ object to control input amplitude as a gain control. Then the audio signal go into rev~2 object. It is the reverb object have four inlet, volume reverb object, feedback(which they call it liveness),  Crossover frequency in Hz and High Freuqnecy damping. When we describe reverb, we will said wet and dry and I am literately using moisture sensor to control it. I am trying to questioning how field recording more realistic and reverb has most acoustic characters in effects. It can recreate an acsotic space. If the conventional way of doing field recording aren’t realistic, then can the space I recreated can be more real in terms of the experience and sensation in the field. So the moisture sensor send signal into the crossover frequency which the more moisture it get, the higher frequency I got in the reverb. The temperature sensor doing the opposite, the hotter it get, more high frequency will be damp and cut out. Then I all go into to writes~ to write files and dac~ to give me an output so I can monitor it while doing recording.

On the other side of the patch loadbang and message are to telling Bela what digital pin I am using. Number 18 is connected to a button which by using select object to trigger the toggle of the recorder. The reason why need a select object is because the button will send out signal when I release it. It would stop the recoding instantly. Number 17 is connected to LED light and it basically function as recording light. It will tell me if it is recoding or not as Bela didn’t have monitor or indicator at all. Finally, I have to copy and create a separate rev2~ patch since Bela didn’t have rev2~ object in Bela pure data.

When I first start the project, I ask Joanne if the Creative Tech Lab have resources that fit my project. After a while, when I start to build and try the hardware part, I borrow one of this sensor. It is DHT11, a temperature and humidity sensor. It sends out both temperature and humidity data in one out put and it is so compact. The hardware seem perfect for my project, however, it took me a week to try to decode the data as it combine both data in one channel. At first, I tried to find any coding sample in GitHub. It is a common sensor and a lot of manufacture produce it, but it is for arduino. All the sample code are for Arduino which I only know so little about it. If I use Arduino just for reading data and send it into Bela, which means I have to bring both Arduino and Bela to the field, plus the recorder as my mic amplifier, it is not efficient at all. Hence, I try to find any sample that can work in Bela, although both is running in same coding language, the software behind is completely different. After a week of doing research and attempt, I still can’t decode it. After chat with Milo and Michael from creative tech lab, I decide to use other sensor that use analog signal.

This is circuit of the hardware for Bela. Firstly, potentiometer connect in analog pin 0. Then the analog temperature sensor connect into analog pin 1 and moisture sensor connected into analog pin 2. The moisture sensor work completely different from DHT11. I measure the moisture of the soil instead of air. For the digital pin, pin 6 connected positive of LED light, and its negative side connected to ground through 10k resister. Pin 7 connected between button negative and 10k resister to ground. Finally, I after I tested it works good, I soldering it into a prefboard which can be more secure and more safe to bring out for field recording.

Another method to record in Pure Data is record through writesf~ object. To do so, it have to send three command into writes~ object, open(create) a wav file, start and stop recording. The problems is if I just use open_[filename], then I would overwrite the last recording. In order to record multiple tracks, I have to make a system that will change file name every time I press open. In order to use one bottom to record and stop (which is the toggle), I used number box and moses object to trigger the record system. When the toggle on, it would be one in number box, and that will trigger the bang in the right to create file and record; when it off, it will trigger the bang in the left and the bang will send signal into message stop. By using integer object to store and recall number and feed +1 object next to it, it will create a loop, and every have bang in, it will +1 to the number that store in integer object. Then the number go into a makefilename object. In the message section, use $ to indicate the various number. As a result, every time when press record, the file name will increase in one from zero. That overcome the overwrite file problem.

As my project is to do recording with Bela, I started to explorer how can record in Pure Data. One of the method is to use the array to capture sound wave. Using ads~ object to convert analog audio signal into digital then use *~ object to control the input signal. In here, I found that can use env~(an envelop follow) with vu meter to track the input signal. Then using tabwirte~ object to send and write data into a table. In order to record one second sound, I have to set the table size into 44100 which is same as the input sample rate (as sample rate in the sample past thought per second).

On the other side, it start with a toggle to run the metro which its speed is control by a Hsilder. Then it send in to a bang that would trigger tabwirte object instantly and send it into a delay object (which I found later that have to type in argument, the patch in above didn’t have delay effect at all) and it trigger the message, and it send 0 and the sample rate and time in to a vline~ object. And the line object send argument into tabread object to play the sound. When I was trying to play this patch with my computer microphone and speaker, it create a feedback loop which create an interesting sound. The speed of the metro affect the sound as it trigger both record and play sample which adding like a gate that chopping the feedback loop. It also create a short attack effect which almost like a kick drum sounds. It was fun, but I found that using this method are good for do short sampling, not suitable for field recording.

At first, when I thought to capture more information like temperature and humanity in the field, my first instinct is build a microphone by changing a mic capsule. Using a mic like objects to do field recording seems to be the straightforward way to do that as field recordist so used to pointing or setting up a mic/mics in the sound source or specific point. However, after I chat with Milo and Joanne, I found that it would be so difficult to make something like that. First of all, mic capsules work different form those sensor. Mic capsules transferring the vibration from the air to electric signal using the magnetic diaphragm, but the sensor is sending out electronic signal. Then I thought, is it possible to use the sensor electronic signal to modulate the audio input (a mic signal). It is possible to do that, but then need a lot of try and error, as Joanne suggested that I can not be done in just one or two months time. Both Milo and Joanne suggested me to use computer like Bela or Arduino. If I use computer, then I can map the sensor data into anything using coding. Working in the way is more accessible and have more control of the data from the sensor. To choose between Arduino and Bela, seems Bela is the better option. As I want those data of the sensor affect the audio in real time, Bela seems is a better choose for processing audio, plus Bela can run PureData which is the only programming language that I have more knowledge with.

To explore more in pure data, I follow the tutorial in archive.flossmanuals.net/pure-data/. In pure data, they only have two type of oscillator, osc~ object create sine wave and phasor~ object create sawtooth wave. Although with some mathematical equation, we can transform those wave into other type of waveform. I was trying to make triangle wave, but the math part really so difficult for me. When I was doing research on how to do that, I found there have another way to create different type of wave which is using array, a table in Pure Data. In the exercise in class, we used it as a wave from monitor to see what wave we have created. However, we can actually read the table and use it as a wavetable synth.

This time using Hslider to control the in put and convert it with mtof object to frequency. Then the function of the phasor~ object is sent out audio signal from 0 to 1 and multiply 2051 which is the length of the table. Then the most important part is tabread object, which can play the table as audio signal. Then I come cross with new box which call Message. Message box can send a number or a command into object box. In the tabread~ object, I send three different message called set waveform, then my pressing it, I can change the tabread object to read different table. The trick of using array is set a common in a message box: first type the table name, then can set it to sinesum with the length and the point. This message can create a desire graphic in the array and use the common normalise 1 to make sure the graph won’t clip and stay inside the table. Then finally sent the tabread signal into *~ 0.5 to low down the volume and sent it into dac~. In addition, I can also use the mouse to draw the graph (waveform array) to create interesting sound as well.

I made my first PD (Pure Data) patch. By doing this, I learned the basic of pure data. Firstly, object: The Rectangle boxes that can type in command and different command works differently. For example, osc~ is oscillator, mtof is midi converter, key is the signal receiver from the keyboard. Object with a ~ signal is the audio signal which have a thicker line with connect to and other object and the thin line is message or number. Object box can also can do math such as * for multiply and it can be use as VCA in pure data to amplify signal. Another type of box is number box. It can show the number of the inlet (signal in) or can change the number in play mode (command + E to toggle edit and play mode). Finally the most important object is the dac~ (digital to analog converter) which basically work as an output.

The patch I made is an additive synth. Firstly used the key object to control changes and used mtof object to convert signal. Then the main first oscillator set in 440 hz. The first number box sent number into four extra oscillator, and use * object to modify the number that can create a harmonic sequence. Then every sub oscillator have different volumes by using *~ object. All the signal go into a *~ 0.25 to limited the final output value to prevent clipping and send that into dac~.