Track 1: Genre Specific Emulation
When this project begun, we decided as a group to choose a song from the ’90’s RnB’ genre option, because it was a style of music that we were all relatively familiar with, but that would also be both challenging and fun to emulate. When decided which song to emulate there were a lot of factors at play that affected our decision. We knew we wanted a song that wouldn’t be too hard to emulate, because we had to make our version sound as close to the original as possible, but we also wanted a song with enough different instrumental and vocal parts in it to fulfil the brief. We eventually landed on ‘Almost Doesn’t Count’ by Brandy. We all agreed that the song was simple enough that we could create a version that would do the original justice, but also involved a lot of work, especially with the vocals, and the mass of vocal layers and harmonies.
As we began planning recording sessions, we knew from the start that we would not need to record drums, and instead would need to program them using samples, because the drums in the original are from a drum machine. This was helpful as it was one less session that needed to be organised. In our first session we recorded Bass and Keys. I organised for a bassist to come in. The bassist was my friend Dani, and in the run up to the session I sent him the material to learn the song, and then organised a time that worked for him to come into the studio. The keys were played by our recording group member Zayn. We decided to record the bass by putting a mic on the amp in the live room, rather than DI-ing it, because the bass in the original version of the song would’ve been recorded the same way. We then also put the keyboard in the live room too, and DI-ed it, so that Dani and Zayn could play through the song together. Instead of playing different sections and overdubbing. We thought it’d be much better for them to just play through the whole song, with the drums and vocals from the original as their backing track. They did a few takes of the song, and once we were happy, we moved on.
Later on in that same session, we recorded the acoustic guitar melody lines player go over the top of the song, and run throughout. This was very tricky because there are some repeated melody lines, but mostly it is seemingly improvised guitar playing, so instead of asking our guitarist to try to learn the entire song and play through it, we went slowly step by step, recording each small melody line one at a time, simply using a singe AKG 414 miked on the 12th fret to capture the performance. I ran the session, and just played each section back to him, repeating sections multiple times so that he could learn that specific section by ear, and then recording in a few takes, before moving onward to the following guitar part. It was a tedious process, but eventually we got enough guitar material that we could then quantise and properly sequence later on.
Our final session involved recording vocals. We knew this would be one of the more complicated session, because the original song, like most 90’s RnB, has many different vocal doubles and harmonies. Nevertheless, we had sent the song to our singer prior to the session so she could learn the lead vocal line, and we recorded that first, recording verses and choruses separately rather than doing full-song takes. Once we were happy with the lead vocal, we moved on to do backing vocals and harmonies. We had the isolated vocal from the original track in the session, and we went through each verse and chorus, listening out for where the harmonies where, finding the notations with the singer, and then layering them over the top of the previously recorded lead vocal. This was a very difficult process, because at times it was difficult to hear the specific notes that each of the harmonies were doing, but eventually we got all of the material recorded, and were very content with the final production.
In post-production, we knew from the outset that the mixing process would be very specific and would require a lot of attention to detail. The general sonic quality of most 90s RnB, including our chosen song, is a very clean and clear mix with a lot of warmth, specifically from the bassline and the vocals, so we tried to achieve that same feeling with our mix, paying close attention to the EQing of our vocals and their relationship to the bass and electric piano, all element that sit in a similar frequency range. Additionally, we knew that getting a clear and pop-y drum sound, so we made sure to use a fast-acting compressor to bring them out in the mix, and to control the high-mids with an EQ to bring out the short and sharp snare drum.
(final emulation track below)
Track 2: Stereo Recording
When tasked with capturing an ensemble solely with stereo pairs and without any individual mics for each instrument, we decided that the best way to do this was to record a string ensemble. My group member Zayn organised 3 violinist and a bassist to come in to record a cover of ‘Howl’s Moving Castle’, and he transcribed the piece and sent it to each of them to learn prior to recording. However, unfortunately our bassist dropped out, so Zayn opted to play piano on the recording in place of the double bass. To get the best capture possible, we opened the upright piano, with the strings facing where we were going to place the microphones, and then sat the three violinists in front of the piano, facing the mics. The stereo mic techniques we chose to use were a spaced pair of AKG C414’s, raised high up and pointing down slightly onto the players. We additionally had an XY pair of KM184s that we placed level with the string players and push them quite close to the violins to get a closer capture and keep them slightly isolated from the piano behind them. The players took around 15-20 minutes of rehearsal, due to the slight change of the arrangement after the piano was brought in, and then we did a few takes of the 3 minute piece until we were happy with what we got. When it came to mixing the performance, the main focus was to get a good blend between the 414 pair and the KM184 pair, as the two different stereo pairs obviously had a very different capture of the same performance, so finding a good balance was crucial.
(final stereo recording below)
Track 3: Remix
When I saw that the brief for the remix section of this assignment included that all drums used in the remix must be found sound, I knew that I wanted to make a dance track, with very over-compressed and distorted drum sound coming from tapping and thudding on different surfaces. I wanted the drums to have an almost ‘dusty’ and ‘grainy’ feeling, taking a lot of inspiration from artists like Burial, Four Tet and Bonobo. The drums was the first thing I create for this remix, and I did this by, first, pointing a microphone at my desk and tapping on it with my hands in a rhythm, making sure to tap my open palm on every second and fourth beat, to act as the snare drum in the song. For some percussion, I took a nearby mug, held it up to the mic, and tapped the bass of the mug on the table, while also tapped my ring on it in rhythm with the previously recorded desk-tapping. Finally, I crucially needed a kick drum, so I took the bass of my fist and slammed it on my desk a few times. Firstly, I EQ-ed out almost all of the high end, and boosted the low end slightly. I then layered different instances of the recorded hit on top of each other and panned them, to give the kick a much wider, fuller feeling. Finally, I grouped the different layers of Kick, and the squeezed them with a fast acting, low threshold, high ratio compressor to ensure that the kick was short and snappy but felt huge.
(video of recording and solo-ed drums audio below)
Next, I needed to build the remix around these drums. I dragged in the acapella vocal from our emulation track, and time stretched it, because the drums I had made were much faster than the BPM of ‘Almost Doesn’t Count’. I then pitched it down two semitones to differentiate it from the original song, and started adding instruments around the vocal and the drums. I knew that I only wanted to use one of two sections of the vocal and repeat them, because dance music often has very repetitive vocal lines. I chose to use the chorus, and after placing it with the drum, I wrote in a bassline that was completely different in notation to the original bassline, but still worked in key with the vocal line. I then built different synth layers on top of that bassline, using its root notes. I then sequenced the track out, creating a buildup, a ‘drop, and a break in a bridge section, and the back into the ‘drop’ before a final outro. I felt that the final drop and outro needed more variation from the first half of the song, so I recorded some guitar and some piano, mainly doubling the synth chord melody. Once that was done, I felt that the remix was finished, had a good sequence, and contained a good use of our original vocal.
(final remix below)
