PJ’s Creative Music Technology Blog

For my interactive Jitter patch I took the concept of music to video and concentrated on that relationship. I wanted to work on the concept of Synchresis by Michel Chion and how we as humans relate sound to image. The basic idea was that within the video there are several events, and each of these events has some sort of sound, ambient and processed, some playing more strongly to one than the other. This meant that each audio event had a relative visual event, and then by categorising each audio event as part of a beat each section could be played back in a rhythmic fashion. This was later added to by the inclusion of the “degrade” and ”jit.sprinkle” objects, both of which degraded image and sound and were used in a way so that when one was activated so was the other, giving a visual representation of what was being heard, degradation.  The interactive process involves the user starting the beat, progressing it, and choosing when the degradation is turned. The beat is progress by cycling through a series of sequences pre-programmed until it gets to the final progression where the beat it produced by a random drum algorithm. The sounds and images are linked by using the frame jump to tool and setting each frame number with a sound. Originally the video had the sounds synched up to each event, but this proved to have timing issues thanks to the video file having to be incredibly small in order to jump to each frame quickly enough. This meant that the audio was encoded in a lesser form, and the lacking of frame rate meant that synch of sound to frame was lost. This was overcome by muting the video and using the trigger to jump to a frame to also trigger a sample, containing the sound at that point. Out of the 2 Jitter based tasks I picked this one to include on the show reel.

I know ive mentioned ths before but this is just an update. Thanks to Edublogs reducing its upload space it means that none of my videos can be embbeded within my posts, so given the time restriants all videos concerning hand ins will be on the disk only, at a later date i will move this blog but for the mean time ill replace all videos with links to somewhere else they can be stored and viewed (most likely youtube).

Edit: I will not be posting links, since the only place i know off hand i could upload to is Youtube, and im sure there is several copyright issues in what i would be uploading.

On completing task 4 and having all of my audio ready to mix I found that there were several problems that would prove as great mixing challenges with the audio. Although the recordings themselves were decent (clean, good levels, no clipping, and good tonality) the performances were lacking. The main problem was timing issues, and in the case of the vocals some pitch issues. My 1^st task on starting task 5 was to isolate and eliminate these problems. Although the original guitar recordings had terrible timing issues I decided to keep them and use them with the 2^nd set of recordings but on having a closer look found that the 1^st recordings were too bad. I still wanted create a multi-track sound from the guitars, so I duplicated the good guitar take 2 more times and just I’d them differently to original, I later panned them hard left and right, made them more top heavy and really pulled them down in the mix. This produced the desired effect. The pitching issues with the vocal were overcome by using the pitch correction object, set at an attack time of 122ms it wasn’t overly noticeable but had a great effect of the sound. I also found that by setting the attack time to 0ms you could get a great sort of robot effect from the vocals, and so used this in the breakdown section of the track. I also added some pitch correction to the guitar solo, since some of the bends were out of key, subtle pitch correction sorted this. The 2^nd vocal track in some places was too similar to the 1^st and I found it created phase cancellation, so to avoid this I added a light chorus to subtly alter the pitch and timing of the track. I then later ran this track through guitar amp pro to give a tinny distorted sound, and put this low in the mix to add to the overall vocal sound. I also used effect achieved from guitar amp pro on the main vocal at one point, to give a large distorted sound. On trying to record a harmony to accompany main vocal it was found that pitching to it was too difficult and so a take was never kept, but on listening to the track I found that it was needed, so I duplicated the main vocal track, added a little chorus to avoid phasing issues and then using the Vocal transformation object increased the pitch by 7 semi tones, giving a perfect 5^th and increased the tonality slightly so that the sound came across as slightly more feminine. I then edited the track so that it only played at certain points in the song. The next main issue was the drums, after sieving through the 4 takes I found that the only way I could make one decent take that sounded good and had no timing issues was to heavily chop up all of the tracks and form one decent track. On beginning this process I found that some sections had issues on all takes, so abandoned this idea and instead imported the decent sequenced drums that were originally used as a guide, and how I composed the drum track. This worked well, so I removed all of the new drum tracks and used and processed the imported one. Once all issues overcome I started to concentrate on producing the track. The new drum track sounded a little thin since none of it had been double tracked but since the whole kit had been bounced to a single stereo track I could double up individual parts, and copying the whole kit gave overpowering high end sounds. To overcome this I copied the whole track and then eq’d it so that there was a low pass filter cutting any sounds above 1 kHz. I then left this new track completely dry added no reverb and brought it down in the mix. I simply copied over all additional eq and effects from the original track.  For the intro I decided I wanted to cut it from the rest of the track, so to start I automated a 2db volume increase on the main output, so the track physically gets louder. I then eq’d the vocals and guitar to give a “telephone” like sound, by adding a low and high cut of around 500hz and 5khz. I then automated these to turn off as when the intro ended, with the high cut being turn off early for the last phrase of the vocal to add to the build. I then ghosted the drums, by cutting the 2^nd part of the intro with the snare rolls and adding this to the intro. I made the send pre fader, turn the fader all the way down and turn up the bus to a reverb, so that only the wet sound would be heard. To keep in fashion with all the other parts of the intro I eq’d this new drum part so that there was no bass end, producing a thin sound. Since the only part left unaltered is the drum part it sounded out of place, despite only being a hi-hit, so I automated a volume change and made the hi-hat allot quieter for this section. .The final main effect I added was achieved by recording the wet signal of the vocals and the guitar solo, off of a bus that had reverb and delay on it. I then added this back into the track and turned in down in the mix. Additional effect to this track included the spreader, since I wanted to make it very distant and ambient, and to just sit in the back ground. I later found that during the guitar solo it could be heard very faintly, and nicely doubled up the sound, so I automated panning on the track for that section so that it would come out on the opposite side to the main lead guitar track. Additional automation includes phasing being added to the guitar for the last section before the breakdown and large amount of reverb on the final hit from the drums for the last part of the intro.

In terms of mixing the piece I mainly used limiters, compressors and reverbs. Some of the track had wild and unwanted dynamics, so I added the limiter 1^st in the signal chain to even the sound out, then the compressor. In most cases I would lightly use the limiter in the compressor to further even out the sound. On the master out I out another limiter, I didn’t wish to squash the sound and lose the dynamics so it’s just very lightly used to add a little more punch to the final mix. I used several different types of reverb, since I wanted each part to sit in its own space, and then parts I wanted to sound together, I used the same bus, i.e., guitars on one, and vocals on another. For the last part of the mix I re-considered levels given the volume changes of the effects added and considered panning. I used panning as the final tool to give each part its own place in the mix, and to give space to the parts that I have sitting in the same reverb, i.e. both guitars for the solo equally panned left and right, but on the same reverb bus. To finish off I checked that all the stems had been topped and tail effectively and cut and lead in and out of the overall track before bouncing the whole mix.

Here the final mix can be heard. (In MP3 format)

Light.mp3

Task 4 and 5 saw required me to make a multi-track band recording and then produce it, concentrating on the mix. The track I decided to record was one written by myself, and to capture it I used myself and friends as performers.  The track listing for the project was as follows;

Rhythm Guitar

Lead Guitar (solo and harmonized solo)

Bass Guitar

Vocal

2^nd Vocal

Drums (including, Kick, Snare (over and under) and Over heads (left and right))

I started by planning how I would record each part and then concentrate of getting s decent clean recording, before considering task 5 and how I would mix it. Each part presented problems that needed to be overcome and issues that I had not previously considered, mainly issues with performers and timing. I started by laying down a simple sequenced drum beat to give a decent tempo and rhythm guide other than that given by the metronome. I then set out to lay down the guitar, since this played out the structure of the song well and could be played with little other pieces. The remainder of the setup was to record a decent sequenced version of the drums and then the guitar again, since there were timing issues and a lack of interest in the sound, this was resolved by being able to play along to the actual drum track. Later the acoustic drums were added when the rest of the track had been formulated around the sequenced drums, since these were guaranteed to be in time.

To record the guitar I captured 4 elements, 2 by microphone and 2 by line. I used a dynamic mic up against the grill slightly off centre of the speaker cone and a large diaphragm condenser mic in line with the speaker but about a metre away. The line recordings were an emulated out from the guitar amp and a DI signal. The photos below show the setup.

I later found that there were timing issues with these recordings, but wanted to keep them for their tonal characteristics, so I re-recorded the guitar (now in time) using a different method and used this as the main sound. This was done by using a Digitec DF-7 distortion pedal, since it produces decent distorted sounds and has nice cabinet and room emulators in it. I recorded this as a stereo signal, since the pedal offered stereo functionality. Photos below show the setup.

To record the lead guitar I used a Line 6 spider amp and the models found emulated in that and recorded straight from the emulated line out. Normally I dislike the sounds produced by Line 6 gear and especially those produced from a line out but this proved to give the bite and ultra distortion sound I wanted for the solo. I still recorded a separate DI signal just in case I changed my mind.

To capture the bass I went straight into the desk via a DI box and recorded 2 signals. The 1^st I left clean and the 2^nd I ran through the bass amp modeller in Logic.

For the vocal and 2^nd vocal I captured 2 takes using 2 mics. The 1^st mic was a Behringer B-2 large diaphragm condenser mic and the 2^nd a very cheap plastic Hitachi dynamic vocal mic. I positioned them so that the B-2 sat in front directly in line with the singer with the dynamic mic just off to the side and behind angled slightly up. I then captured 2 decent takes of the vocalist, and kept one for the B-2 and the other for the dynamic. This meant that I had 2 different takes with 2 different tonalities playing alongside each other. The B-2 produced a clean present sound that you would expect from a LDM while the dynamic mic produced a more trashy sound, this was later processed accordingly. Photos below show the setup.

The initial plan was to use 2 small diaphragm condenser Octava mics on the snare, one directly under and one slightly off axis over both with 10db pads. Unfortunately the mic clips proved to slack to hold the mics at the appropriate angle, so I swapped for a couple of Shure sm-57’s both with sturdy mic clips. I later found that thanks to technical issues I could only record 4 channels simultaneously thanks to problems with the Behringer interface which provided any over 4. This meant I could only record 4 tracks, and since there were 5 planned tracks I had to eliminate one, and picked the snare under, leaving me with Kick, 2 over heads, and snare over. With this in mind I knew I would lose a lot of the snares tonal character so positioned the OH’s appropriately to capture more of the snare accurately. I did this by making sure both capsules were exactly 120cm from the centre of the snare to avoid and phase issues (this also proved necessary since the snare was by far the loudest drum and would be captured by all mics.) and angled the mics so that they were off axis from the snare so that the attack of the snare did not over power what I wanted the OH to capture. This also proved to produce even ambient cymbal sounds, but unfortunately took away from the tom sounds, which ended up being far too quiet and lacking body.  For the overheads I used 2 SE Electronics LDM’s and positioned them as mentioned above, with each concentrated on half the kit. The left designed to capture hi-hat, snare, crash and 1^st tom, whilst the 2^nd captured 2^nd and 3^rd tom and the ride. This effectively cut the kit in half but positioned the right OH (as if you’re the drummer) slightly further towards the drummer so that the centre line through the kit was off at a 45 degree angle. To record the kick I used a Shure sm-58 due to a lack of a dedicated kick mic. I then used the goose neck extension for the mic stands due to a lack of a short kick stand and put a guitar pedal in the kick drum to angle the mic up at the place where the beater hit due to the goose neck being impossible to position  effectively, and refusing to stay in the place. A large pillow absorbed any sound travelling from the body of the kick through the pedal to the casing of the mic. Although very precarious the mic stayed in place and produced a decent sound, with a little snap. I swapped the kick pedal for my own, which not only did I know was not broken, but had the option for a plastic beater head. The plastic head produced a sound with far more snap than that produced by a felt head, which gives more of a thud sound. To record the snare, now only using 1 mic I angled the sm-57 towards the centre of the drum but just off the point of impact. I then positioned the mic slightly forward of the hi-hat so that it did not become over powering and far enough back so that it did not get hit with a drum stick. To finish off I taped small pieces of foam to each of the drum heads to help eliminate nasty overtones and slackened the rear skin of the kick drum to deaden overtones produced there. Photos below show the drum setup.

For this task it saw us recording 3 different natural instruments with 3 different stereo mic techniques. I picked AB Stereo as a standard and DIN Stereo and a modified version of OCT Front System as the other 2. The 3 instruments used where Flute, Acoustic Guitar and Cello. For the recordings i used a stereo matched pair of small diaphragm condenser mics called T.bone em700’s, and a Behringer B-2 large diaphragm condenser mic. I used the inbuilt 75hz roll off on the T.bones and a 75hz roll off on the desk for the B-2 since i didnt know the exact frequency point on the mics built in roll off. The T.bones where set in a Super Cardioid pattern for all recordings, and the B-2 in a cardioid pattern. For the initial setup i setup all the mics with phantom power on, and listening over headphones adjusted the gain on the desks pots for the best signal to noise ration, by increasing gain until gain is audibly heard and then backing off a little. I then recorded post fader on the desk and used the faders to control the input into Logic. Whilst recording i kept checking the signal on the pots using the PFL to check for any clipping, but none ever occurred.

For the AB Recordings i used a spacing of 80cm for all recordings and simply altered the height for each instrument. Moving the mics closer limited the stereo effect, ambiance and removed lower tones. Although not so relevant for the flute i wanted to keep the spacing consistent.

I then continued to use the same technique for the guitar and cello. I found that A-B produced the best sounds for the flute and guitar. Although they where not as full sounding as the DIN technique the slightly toppy sound suited the flute perfectly, and produced a very clean sound for the guitar with nice but to over-bearing pic noise. The cello was also nice and clean, but came across as far to distant and spaced, and didnt accurately represent the sound of the actual cello.

For the DIN Stereo recordings i angled the mics 90 degrees away from each other with the ends of the capsules being 20cm apart. This technique produced far more fuller sounds unlike AB, but sacrificed stereo image, giving a more concentrated sound. For the guitar DIN gave a nice sound, but very different to that produced from AB, i wouldn’t say it was worse, but just gave a different tonality, one that was far fuller with greater dynamic range. Initial tests gave a sound with far too much pic noise so i kept moving the mics until i got a placement that eliminated this issue. I also experienced a similar issue with the flute and keynoise. Eventually i found that by positioning the mics slightly off axis i could still get a great direct sound but without the unwanted noise. The photos below show the initial setup, and then the altered one used for the recordings.

The DIN technique didn’t seem to suit the flute well, the fuller sound and lower tones made the flute sound very flabby and lost all the clarity AB had offered. Despite the concentrated sound experience from DIN so far the cello sounded oddly ambient and distant. Given my predictions based on the guitar and flute sound i thought it would work well with the cello sound, but again the cello i felt wasn’t accurately represented by the technique. This is possibly due to me having to position the mics quite a distance from the player, and a lot higher than wanted, but this was necessary to eliminate yet another unwanted player noise, bow friction.

Overall it would seem that DIN produced a decent alternate sound for the acoustic guitar, that could potentially be used if the guitar were to be heard by itself, but if i was mixing it with other instruments i would use AB, but wasn’t very successful for cello and flute mainly because of its amazing abilty to easily pick every unwanted sound from each instrument, leading to sacrificed mic placement.

For my final 3 mic technique i used a modified version of the OCT Front System by stealing elements from the Decca Tree technique. I kept everything about the OCT technique the same including the mic pickup patterns (super cardioid on the stereo pair, and cardioid on the centre mic), except i angled the stereo mics forward like in a Decca tree layout, instead of them facing outwards away from each other. I did this because of the early recordings showing that AB produced great clear results, but guitar and cello lacked some depth, something i hoped would be helped by the additional centre mic. Again like AB i kept the spacing between the stereo mics at 80cm and only altered the height for different instruments. The centre mic i positioned on the same plain as the stereo pair, only 8cm further forward, and directly in the middle between the 2. The technique looked like as follows.

 The modified OCT technique produced varied results. The additional centre mic picked up the flutes key noise in all its glory and i found this to be unavoidable. The guitar sounded very boxy, again i guess the additional mic exaggerated lower tones from the guitar, and the higher tones being slightly quieter thanks to the mic placement sounded a little lost. The cello of the other hand i felt was represented perfectly by the modified OCT system.  My initial thought was that the centre mic would pick up all the bow noise, but by positioning the mics slightly higher than the cello’s body this was completely avoided and instead a true and accurate sound of the cello was captured. The additional mic also helped capture the cellos large dynamic range, by concentrating on that and the more concentrated tones, whilst leaving the stereo mics to the ambiance of the room, which would carry less dynamic accuracy.

Overall i feel the recordings went very well and each technique produced interesting results, and demonstrated which techniques suit different instruments. Although problems where found these where eaily overcome by experimentation and using previous knowledge. Issues include a very bad signal to noise ratio on the left stereo mic, which unfortunaltly left a low noise hum on a couple of test recordings, but was neccesary to get levels high enough. A quick checked showed that it was a bad connection in a lead that was producing the hum, so recordings where shorly held up whilst i took the lead apart and re-soldered the connections, eleminating the hum.

I used the following resources in gathering information of mic techniques:

A-B Stereo:

http://www.dpamicrophones.com/en/Microphone-University/Stereo-Techniques/AB-Stereo.aspx

http://www.schoeps.de/PDFs/stereo-recording-techniques-e.pdf

DIN Stereo;

http://www.dpamicrophones.com/en/Microphone-University/Stereo-Techniques/DIN%20Stereo.aspx

Modified OCT Front System:

http://www.schoeps.de/PDFs/oct-e.pdf

http://www.dpamicrophones.com/en/Microphone-University/Stereo-Techniques/Decca-Tree.aspx

I read and heard that Edublogs has imposed a tiny limit on the amount of media you can upload, 20mb if i remember. Having searched i can seem to find where this was stated again, so for ythe mean time ill be staying at edublogs, unless it means i cannot upload my task, in which case i will move my blog and state here where to. Thanks.

Strangely creating the DVD show reel proved to possibly be the hardest task. Constant technical issues with finding an appropriate software, appropriate software’s proving to not work, and file type and conversion issues meant it took a large amount of time to finally create the DVD. Eventually it was created using Windows DVD Maker (only available on Vista, which I didn’t have! And iDVD refused to work before people ask why I didn’t use that) and the films are Mpeg2 (mpg) files. Unfortunately due to the limitations of the software I couldn’t include any additional files to the dvd, so the dvd is just the show reel and all other work are on an additional CD. The menu music is hard drive interference that I was struggling to get rid of, so decided to record it since I liked the sound (except when it’s not wanted) before eliminating it from the system by removing an earth loop. On testing the DVD I found that for an unknown reason the ends of the videos pause, but the audio continues. I tested it on a PS3 so I doubt its a player issue, probably due to bad encoding by Windows DVD maker. All dvd content is available on the additional CD, so if issues persist the original versions are available. The three tasks chosen were 1, 2 and 8, Alternate soundtrack to Bergman’s Persona, Scoring for an Animation and an Interactive Jitter patch.

For the 1st of the Jitter tasks I considered the direct relationship between colour and sound. Initially my idea was to represent each audible colouring of a three part chord, with actual colours, i.e. sad chords such as minor to have dull murky colours. But this was found to be far too vague, and peoples individual ideas of how a chord should look in terms of three colours seemed to varied and far too specific to b considered. I altered the direction of the patch and now every time its clicked a random chord is generated (within traditional musical rules, i.e., major, minor, augmented, diminished) and in turn a 3 part image is produced, each part representing that note of chord in relation to the other notes of the chord. This was originally going to be my Jitter task for the show reel, but found that without further research it didn’t have the body to achieve what I wanted it to.

Musical interpretation of a Painting.

To be uploaded.

Task 5 saw me recreating the sound design for the sequence of a film and acting as a foley artist. Initially I wanted to do a sequence from Star Wars, but the group I was working with decided the opening scene from Gladiator. After a few weeks of not much happening I decided to leave the group and go back to my original plan of Star Wars. Since my earliest memories of watching Star Wars I remember loving the sounds that Ben Burtt invented for all of the fictitious elements, my favorites being those R2-D2 made and of course Light Sabers. At first I considered doing a scene with R2-D2 in it, but most had dialogue, something I wished to avoid, plus I soon came to realize just how hard it would be to re-create the sounds made by R2-D2 so chose a Light Saber battle instead. Conveniently I found a re-cut version of a fight scene from episode one that removed all dialogue and concentrated on the fight so I used this. Now I know that the sound Ben Burtt created are fictitious, so when designing a sound for R2-D2 or the Light Saber there is no need to try and emulate the originals, but I soon found that Ben Burtt’s versions pretty much hit the nail on the head, and what with Star Wars being so legendary although the Light Saber’s is fictitious it has a sound that is set in stone. So my first task, how to make a light saber sound? My basic understanding of physics told me that the “whom!” sound produced was thanks to the Doppler effect, the idea that as sound waves contract (move towards you) they go up in pitch, and as they stretch (move away) they go down in pitch, it’s the ambulance siren analogy. The next issue was how to create this, and then how to get it to sound like the images of the film. Some quick research showed me how Ben Burtt did it and I went about replicating this. First I created a low noisy hum; I did this by recording an unbalanced jack lead that I shorted the end out with by grabbing it in my hand. I processed this a little and created a loop so I had a constant nice hum. The next part is the clever bit, I then blasted this sound out of a single speaker and recorded it again with a microphone, only I held and moved the microphone like the actors moved their light sabers. This meant that as the mic moved across and towards the speaker the Doppler Effect happened, moving the pitch up and down, what’s more is that it also varied the dynamic, increasing as the mic got closer and on axis with the speaker. First problems where with wind noise being picked up thanks to the mic moving through the air, but thus was soon go around by moving the mic as gently as possible and using a foam wind shield around the capsule. I then recorded 2 takes, one in which I, to the best of my abilities, tried to recreate what the actors were doing with their sabers in term of movement (in the scene there are 3 sabers, 4 if you include the double ended one, so I concentrated on replicating whichever the camera concentrated on) and the 2^nd producing the underlining idle hum of the sabers by moving the mic slowly closer and further from the speaker. With a combination of repositioning m live take and cutting and pasting element of it I was able to recreate the sounds of the saber movement in an organic and real fashion. The 2^nd major sound of the light saber is that made when they hit each other, ranging from a series of blasts as they bounce off each other to large crackles as they slide against each other. Ben Burtt’s sound was very electrical with elements of clashing metal. I knew I could get the electrical sound by shorting out my car battery since it produces a great spark and noise little a small thunder strike (don’t ask how I know this) but I didn’t fancy doing this again so set about getting a similar sound other ways. The sound was found in a small bottle of water I keep beside my desk. By violently shaking it in single strokes you got the impact of the water against the side of the bottle and then some splashing sounds after. I recorded several of these so that the same sound wasn’t used for every hit, and then set about processing them. I increased the pitch by a few semi tones and then time stretched them until I got that great grainy quality. All that was left was to position each of the hits, this was relatively easy since every hit was accompanied by a huge light bloom so I just had to scroll through the film finding each blast of light, and then appropriately position the sample. In some cases the sabers would touch and stay together for a short period of time, so a blast sound wouldn’t work. To replicate this I sounded the initial hit copied it next to that sound file then reversed it and stretched it for the time of contact. With the main and important elements complete all that was left where any additional foley such as the explosion of the door controls, the box that hit the door moving and all the actors movements (footsteps, kicks bashing).