Hands on?

You’ll need:

• My hands-on files
• Audacity (Linux, Mac, or Windows)
• Max (free Runtime version)
  • Mac OS X
  • Windows

Sound basics

• Sound is motion…moving back and forth…vibration…patterns of vibration, really.
  • Waveforms, timbre
  • Limits of hearing pitch
  • Sample rate
  • Bit depth
  • FFT
• …but we don’t always see the pattern. We think at the control rate (1–1000 times per second) but we sense at the audio rate (e.g., 44,100 times for second for CD-quality).

Audio editing

• Fade edits to avoid clicks—audio rate! LIsten to the clicks at the beginning and end of the music clips. Use the Fade In and Fade Out features to remove the clicks. Even if you want a “sudden” cutoff, 20ms or so will still feel sudden but won’t click, betraying amateur editing chops. This feature should be the same in any audio workstation.
• Ambient noise—audio rate!
  • Mic distance matters, room treatment helps too. Listen to the three different vocal tracks and note the ambient noise. One is overdriving the input so the tops and bottoms of the waves are cut off (zoom in to see). This is very difficult to fix; it’s almost always better to re-record it.
  • Noise reduction techniques
    • Try the Equalization process to cut out the background hiss/rumble. Can you get rid of the noise without taking away from the vocal? This feature should be the same in any audio workstation.
    • Noise gate: Audacity doesn’t come with this, but it’s a kind of automatic volume control. It cuts out all sound if it falls below a certain threshold. This doesn’t involve the sacrifice in timbre involved with Equalization, but it can be disturbing for the ambient sound to jump in and out.
    • Try the Noise Removal feature. This is FFT-based noise reduction. Use a “silent” bit at the end of the vocal clip to capture the “signature” of the noise, then highlight the whole clip to remove that noise from it. Try it at extreme values of reduction to get familiar with the chirpy/warbly artifacts of FFT-based (frequency domain) processes. Now undo that and try to do it again more subtly.
    • Try adding reverb or keep the music in, mixed low, to cover up the ambient noise.
• Balance—control rate. Try these different approaches, one at a time, to balance the music tracks and a vocal track. Each can be an effective way to achieve balance. The key is to find some “space” for each important sound to have to itself, whether in the frequency spectrum, stereo panorama, in time, etc.
  • Volume fader. This feature should be the same in any audio workstation.
  • Panning left or right. Keep important things near center (direct center isn’t always the best). For a vivid texture, pan contrasting rhythmic tracks far left and right. To let elements “blend” into the background, put them closer to 45º left or right. Since the music tracks were already a full stereo mix, try panning them hard left and right and see how it affects balance with the vocal. This feature should be the same in any audio workstation.
  • Peak or notch EQ. In Audacity, try the Plot Spectrum feature on a vocal clip. Notice it has peaks around 450Hz (where the fundamental pitch of the voice is) and 2500Hz (where we hear the difference between vowels). Instead of boosting the vocal track, try cutting the music tracks at this frequency to carve out a space for the voice to come through. Other programs may not have visual tools like this but may let you “sweep” a peak/notch filter through the frequency spectrum. Boost and sweep through the track that needs clarification until you find parts of its sound that are interesting or crucial to intelligibility, then undo that filter and cut the music at the same frequency as described  above. Cut instead of boosting whenever you can.
  • Automate volume. Use the Envelope Tool in Audacity to draw curves that dictate how the volume should change over time, to fade the music out when the voice enters. You could also use this to briefly reduce the sound if one note is covering up a word.
  • Ducker. Audacity doesn’t come with one, but this is an automated dynamics process (related to the gate) that is basically a compressor on a sidechain. A compressor doesn’t silence sound, but it turns it down temporarily when it goes over a certain threshold, them turns it back up (like a nervous parent watching the stereo at a kid’s party). The sidechain means the you could make the compressor listen to the voice and change the volume of the background music. That’s a ducker.

Signal processing

• Types
  • Dynamics—gate, compressor, limiter, expander
  • Filters/EQ
  • Delay based effects—echo, reverb, chorus, flange, sometimes pitch or speed change
  • Frequency domain effects—for noise reduction, sometimes for pitch or speed change, more unusual effects
• Introduction to the Max or PD graphic programming environment
• Most effects are delay based, even filters
• Frequency domain effects are powerful

Conclusions

• When investing in your studio, it’s worth paying more for analog gear (mics, preamps, Analog to digital converters (ADCs), less so for DACS and speakers), complex algorithms (e.g., reverbs), FFT-based effects, and features like parameter automation.
• Tools make assumptions for you; it makes some things difficult/impossible as it makes other things easier. Always be aware of this, and find a tool set that meets your functional and creative needs. Check out the interactive DSP Max patch. Open a sound file and try activating each of the “smart” effects processes that change the sound based on properties of the incoming sound, e.g. pitch, loudness, brightness, noisiness, etc.