How sound works
You don't need to be a sound engineer to record and edit your voice well, but you do need to understand four ideas: what a wave is, what frequency and volume are, how decibels are measured and what the numbers 48 kHz and 24 bits mean. With that, everything else in the course will make sense.
Sound is a pressure wave
When you speak, your vocal cords vibrate and push the air. That air compresses and expands in waves that travel to an ear or a microphone. The microphone turns those pressure changes into an electrical signal, and the recorder turns it into numbers. That's the whole chain: vibration, air, microphone, signal, file.
We mainly care about two things in a wave: how many times it vibrates per second (frequency) and how strongly it does so (amplitude). In the simulator above you can move both and see how pitch and level change.
Frequency and pitch
Frequency is measured in hertz (Hz): vibrations per second. The more there are, the higher the sound. At best, the human ear hears from about 20 Hz to about 20 kHz (20,000 Hz), and some of the highs are lost with age.
When working, it helps to split that range into zones, because each one sounds different:
| Zone | Approximate range | What it sounds like |
|---|---|---|
| Sub-bass | 20–60 Hz | Rumble you feel more than hear; in a voice recording it's usually noise (bumps, traffic) |
| Bass | 60–250 Hz | Body and warmth |
| Low mids | 250–500 Hz | Too much and the voice sounds “muddy” or “boxy” |
| Mids | 500 Hz–2 kHz | The core of the voice and of intelligibility |
| Upper mids (presence) | 2–5 kHz | Clarity: the voice “steps forward” |
| Highs | 5–10 kHz | Brightness; this is where hissing S sounds live |
| Air | 10–20 kHz | A sense of openness and detail |
The range of the voice
A voice isn't a single frequency. It has a fundamental frequency, which is the pitch you speak at, and many harmonics, multiples of that fundamental that give it its own timbre. That's why two people saying the same word at the same pitch sound different.
- Fundamental of lower voices: roughly
85–180 Hz. - Fundamental of higher voices: roughly
165–255 Hz. - Presence and intelligibility: around
2–5 kHz, where consonants become clear. - Sibilance: S and Z sounds are concentrated between
5–8 kHz.
These numbers will come back in EQ: you'll know what to adjust when a voice sounds dull, muddy or has piercing S sounds.
Since the lowest fundamental sits around 80 Hz, whatever is below it is usually noise: bumps on the desk, traffic, air conditioning. That's why it's so common to apply a filter that cuts that zone.
Amplitude, volume and decibels
Amplitude is the strength of the wave, and we perceive it as volume. It's measured in decibels (dB), a logarithmic scale: it doesn't add, it multiplies. As a rule of thumb, +6 dB is roughly double the signal's amplitude, and it takes about +10 dB for something to seem “twice as loud” to us.
The problem is that “decibel” is used for different things. These are the two that matter to you:
| Scale | What it measures | Reference | Example |
|---|---|---|---|
| dB SPL | Sound pressure in the air, what reaches the ear | 0 dB SPL is the threshold of hearing | Conversation at one meter: about 60 dB SPL |
| dBFS | Signal level inside the digital file | 0 dBFS is the maximum possible | Peaks of a well-recorded voice: between −12 and −6 dBFS |
In dBFS the values are negative: the closer to 0, the louder. And 0 dBFS is a hard ceiling: if the signal hits it, it gets cut off and distorts. You'll go into this in depth in levels and gain.
Mono and stereo
A mono file has one channel: the same sound comes out of both speakers. A stereo file has two different channels, left and right, which create a sense of space.
A voice recorded with one microphone is mono by nature. The norm in podcasting is to record each voice on its own mono track and, at the end, export in stereo if there's music or effects with space, or in mono if it's voice only. Panning each speaker hard to one side is uncomfortable on headphones: keep voices in the center or close to it.
If you record a microphone onto a stereo track with the input on only one side, the voice will play in only one ear. Check it on your first test: it's one of the most common beginner mistakes.
Sample rate and bit depth
To digitize sound, the recorder takes “snapshots” of the wave many times per second. That's the sample rate. And each snapshot is stored with a certain precision: the bit depth.
| Setting | Common values | What it means | Recommendation |
|---|---|---|---|
| Sample rate | 44.1 kHz / 48 kHz | Samples per second; captures up to half that value (about 22–24 kHz) | 48 kHz, the video standard; 44.1 kHz works too |
| Bit depth | 16 / 24 bits (and 32-bit float) | Precision of each sample and the margin between noise and the ceiling | Record at 24 bits; export according to the platform |
Going above 48 kHz won't gain you anything audible for a voice, and the files will be bigger. What matters is not mixing sample rates in one project: choose one and keep it in the recorder, the editing software and the export.
24 bits won't make the voice sound “better”, but it gives you a lot of headroom: you can record at a slightly lower level without noise appearing when you turn it up later. That peace of mind is worth a lot when you record several people who don't control their volume.
- Frequency (Hz) is pitch; amplitude is volume.
- The voice has its fundamental between about 85 and 255 Hz, presence at 2–5 kHz and S sounds at 5–8 kHz.
- dB SPL measures sound in the air; dBFS measures the digital signal, with 0 dBFS as the ceiling.
- Each voice on its own mono track.
- Record at 48 kHz and 24 bits, and don't mix sample rates in one project.
Your turn: put numbers on your voice
Record your voice and find out which frequency zone and level it falls into.
- Set your software or app to 48 kHz and 24 bits and record 30 seconds of normal speech.
- Look at the meter: note the dBFS range your peaks move in.
- Open a spectrum analyzer (Audacity has a free one) and find where most of your voice's energy is.
Frequently asked questions
Which is better for a podcast, 44.1 kHz or 48 kHz?
Both work for voice. 48 kHz is the video standard, so it's the most convenient choice if you'll ever publish on YouTube; what matters is using the same one throughout the project.
What's the difference between dB and dBFS?
dB SPL measures sound pressure in the air, with positive values. dBFS measures the level inside a digital file, with negative values and a maximum of 0 dBFS.
Should I record my podcast in mono or stereo?
Record each voice on a mono track. When exporting, use stereo if there's music or effects with space, and mono if there's only voice.