Frequently asked questions

This is the first question almost everyone asks when they decide that they want to create audio or video content. And the answer depends on a lot of things, but for video, I usually recommend a quality boom microphone for indoor recording or a shotgun microphone with a wind shield for outdoor recording. Here's a video which goes into more detail on choosing a shotgun microphone. And these are the microphones I use most often in my work.

Both can work well. Condenser mics usually need 48V phantom power and tend to have a more extended, brighter high end. I prefer them when the room allows. Dynamic mics are often recommended for untreated rooms, but not because they magically reject noise. They're less sensitive, so people work right up on them, and that improves the signal-to-noise ratio. In a noisy or untreated room, a dynamic mic used close is a safe bet. More in this video on microphone specifications.

You can, but it's not ideal. The interference tube that makes a shotgun directional works well outdoors. Indoors, though, sound bounces off walls, ceilings and tables into the side of the tube, which colors the voice and makes it sound less natural. For booming indoors, a supercardioid pencil (small-diaphragm) condenser usually sounds more true to life. My favorite is the Sennheiser MKH50. If you can only afford one mic and you mostly shoot outdoors, a shotgun is still a sensible choice. More in this video.

Only if your preamp can't supply enough clean gain. The SM7B needs a lot, so look for at least 60–65 dB of clean gain. Many modern interfaces, mixers and recorders have plenty; the Mackie DLZ Creator XS, for example, offers up to 80 dB. Sometimes an inline booster actually makes things worse. With the RØDECaster Pro II, adding one can raise your overall noise floor. Check your preamp's specs before buying one. More in this video.

Closer to the person talking than you think. The farther away the mic is, the more room sound it picks up and the thinner the voice sounds. A mic on top of a camera two or three meters away will almost always disappoint. For talking-head videos, I boom a mic just out of frame, about 30–40 cm from the mouth. A podcast-style mic should be closer still. Moving the mic closer improves your signal-to-noise ratio more than almost any gear upgrade. More in this video.

Adjust the mic position before reaching for plugins. With a large-diaphragm condenser, putting the capsule directly in line with your mouth tends to emphasize sibilance. Moving it slightly below or above mouth level smooths it out. A foam windscreen also softens the high frequencies a bit. If you still need help in post, a de-esser or a tool like Sonnox Soften works well. Don't overdo it, though, or your S's start to sound like a lisp. More in this video on mic placement.

Clothing rustle is the biggest challenge with lavs.

• T-shirt: turn it inside out and tape the mic at the neckline under the ribbing. Then run the cable along the seams with 3M Transpore medical tape.
• Polo or button-up: attach the mic with double-sided "stickies" near the bottom of the button placket. Put a small piece of fur over the grille without covering it. Add a cable saver about 15 cm down the cable so tugs on the cable don't reach the capsule.

The fur makes the biggest difference. See how to do it under a t-shirt and under a polo shirt.

A lav sits on the chest, so it sounds more attached to the body and less open than a boom mic above the talent. That can be jarring if you need to cut between them. EQ matching helps: set the boom as the reference and let an EQ match tool apply a corrective curve to the lav. DaVinci Resolve's Fairlight has a built-in EQ Matcher. In my tests, though, FabFilter Pro-Q 4's EQ Match produced a much closer, more convincing result. More in this video.

Wind hitting the capsule directly causes distortion you usually can't fix in post, so plan for it.

• Foam covers handle only the lightest breezes, around 2 mph, and noticeably dull the sound.
• Foam-with-fur covers handle a bit more, roughly 5–6 mph.
• Blimp or Zeppelin-style windshields handle the most wind. With those, I go without the fur up to about 4–5 mph and add the fur above that.

More in this video.

Not really. In my tests there was no measurable difference in noise floor between cables. Pricier cables are mainly better shielded against interference. How you run them matters more:

• Keep audio cables away from power adapters.
• Cross power cables at 90 degrees.
• Never bundle audio and power cables together.

A cheap cable lying next to a power brick is a classic cause of buzz. My recommendation is Canare or Mogami cable with Neutrik connectors. More in this video.

Use directional microphones to focus on your subject and minimize ambient sound. Record in quiet environments, and consider using sound blankets to absorb high frequency reflected sound. Post-production noise reduction tools can further reduce noise. Keep in mind that AI audio tools cannot make a really, really bad audio recording sound great. It is always best to start with the cleanest recording possible. And if you can only do two things: 1) sound blankets, 2) turn off noise generating things when recording.

For most video projects, record audio in uncompressed WAV format at 48kHz for optimal quality. I'd suggest you avoid compressed formats like MP3 for recording. Compressed formats like MP3 and AAC are best for final delivery after you've post processed and sweetened the audio.

A good 32-bit float recorder feeds your mic into two or more analog-to-digital converters at fixed gain. One handles quiet sounds and another handles loud ones, and the recorder combines them into a file with enormous dynamic range. If a recording goes over 0 dB, you can usually turn it down in post and recover it cleanly. That removes a lot of stress for solo shooters.

It doesn't make you immune to problems, though. Your mic can still distort on very loud sources. For live streaming, the signal is converted down before viewers hear it, so you still need to know how to set gain properly. More in this video.

No. The format alone doesn't guarantee clean audio. Budget recorders often use a single converter and cheaper preamps, so when you boost quiet dialogue in post, the noise floor comes up with it. How well a recorder recombines its converters also matters; poor recombination can sound harsh. In my testing, these sounded genuinely good:

• Sound Devices MixPre series
• Zoom F3, F6 and F8n Pro
• Zoom H5 Studio

Listen to unprocessed samples before you buy. More in part 1 and part 2.

Probably not. Modern preamps are so clean that your mic choice, mic placement and room noise have a much bigger effect than the recorder. A Zoom F6 or Sound Devices MixPre is more than good enough for dialogue. Upgrade when you need specific features, such as:

• more inputs
• more flexible routing
• timecode
• balanced outputs for paid production work

Don't upgrade just hoping for better sound. More in this video.

Consumer 2.4 GHz kits, like the RØDE Wireless GO/Pro, DJI Mic and Hollyland Lark, are great for simple one- or two-person setups. They're small, easy and affordable. But they share the band with Wi-Fi, so they can struggle in crowded RF environments, and you can't retune around interference. UHF systems let you scan for clear frequencies and generally get through walls better. They also scale to many more channels, which matters on bigger productions. More in this video.

Line of sight matters more than the range spec on the box. In my tests, a system that reached 100 meters in clear line of sight dropped out at about 22 meters when the transmitter faced away behind the talent's body. Keep a clear path between transmitter and receiver.

On systems with removable (SMA) antennas, you can put an external antenna on a cable. That lets you raise it above obstructions or move it closer to the action. The benefit comes from that positioning, not from the antenna itself. More in this video.

Use a clapperboard or hand clap at the start of each take to create a visual and audio sync point. In post-production, align the audio waveform with the visual cue. Many video editing apps have automatic sync features which use the audio in your video clip to sync the audio from your audio recording and they work quite well. We call this dual-system sync sound.

Timecode stamps each recording with a precise time (hours:minutes:seconds:frames), so your editing app can line clips up automatically. Camera and recorder clocks drift, so you'll want external timecode generators with accurate temperature-compensated clocks.

• Jam-sync the generators together and leave one connected to each device for the whole shoot. If you disconnect one, that device falls back to its own drifting clock.
• Even with timecode, very long takes can drift a few frames by the end, because timecode only marks the start of the clip.

For many creators, waveform-based auto sync works fine. Timecode shines with multiple cameras and long shoot days. More in this video.

YouTube allows up to −14 LUFS, but for dialogue-heavy videos I target around −17 LUFS. −14 suits pop music but gets fatiguing over long stretches of talking. Always leave at least 1.5 dB of true-peak headroom, because the AAC or MP3 encoding on export eats into it. More in this video.

Level first, compress second. Pushing raw dialogue straight to a loudness target with heavy compression sounds unnatural and in-your-face.

1. Run a dialogue leveler first, with a gentle range of around 3 dB.
2. Add a light compressor with its threshold set just above the remaining peaks, roughly 4:1 with a fast attack, so it catches only what sticks out.

If you use DaVinci Resolve's built-in dialogue leveler, turn off "Lift Soft Dialogue," which mostly just boosts noise and breaths. More in this video.

Reduce breaths rather than cutting them out completely; removing them entirely sounds fake. Breath tools have improved a lot. iZotope RX 12's Breath Control reliably finds breaths without grabbing consonants the way older versions did. I use its target mode to bring breaths down to around −50 to −55 dB, with "natural" reduction so the room tone isn't gated. More in this video.

That "echo" is actually reverberation: sound bouncing off hard, flat surfaces. The best fix happens before you record. Hang sound blankets around you and put one on the table in front of you; any thick blanket or old duvet will help. If you're stuck with a reverberant recording, de-reverb plugins like Waves Clarity Vx DeReverb or iZotope RX can help a lot, but they won't make a bad room sound perfect. More in this video.

Keep the music noticeably below the voice, and carve out room for speech.

• EQ cut: cut the music in the range where the voice sits, so the music stays audible and the dialogue stays clear.
• Automation: automate the music's level and EQ so it's fuller during the intro and backs off when talking starts.
• Sidechain compression: have the music dip automatically whenever someone speaks.

More in this session on mixing music behind dialogue.

Yes. Auphonic will level, EQ, reduce noise and loudness-normalize an exported mix, and its free tier covers two hours of audio a month. Plugins like Accentize DialogueEnhance and Noiseworks Dynassist can also do a lot inside your editor. Two cautions:

• Don't use automatic silence cutting on video, because it breaks sync with picture.
• Set noise and breath reduction by ear rather than on Auto, which can sound robotic.

More in this video.

Usually it's mic bleed. When several mics are open in the same room, each one picks up everyone's voice a split second late, which causes comb filtering. More open mics also means more room noise. To fix it:

• Give each person their own mic and have them work close, within about 10 cm.
• Use cardioid mics and seat people so they're behind each other's mics.
• Spread people out.

Then either ride the faders, cut out the non-talking sections in post, or use an automixer. Automixers are built into the Sound Devices MixPre, Zoom F6 and F8n Pro, and some podcast mixers, and are also available as plugins like WT AutoMixer. More in this video.

Audio usually arrives before video from a capture card. To measure the offset:

1. Record a short clip in OBS while clapping a few times.
2. Open the clip in a video editor and step through frame by frame.
3. Count how many frames the clap sound comes before the visible clap.
4. Multiply that number by the frame length: 33.33 ms at 30 fps, or 16.67 ms at 60 fps.
5. Enter the result as the sync offset in OBS's Advanced Audio Properties.
6. Record another clip to confirm the fix.

More in this video.

There are two common routes:

• USB into your computer. This is the easiest, but you'll probably need to delay the audio to match the video.
• Into your camera. If your mixer has balanced outputs and your camera has XLR inputs, a balanced cable does the job. If your camera only has a 3.5 mm unbalanced input, use a balanced-to-unbalanced converter like the ART CleanBox Pro.

More in this video.

Yes, with some caveats. Headphones remove the room and the natural crosstalk between your ears, so mixes don't always translate to other playback systems. A couple of tools help: IK Multimedia's ARC On-Ear and Beyerdynamic's Headphone Lab plugin (free for Beyerdynamic owners) apply correction and simulate listening to speakers in a room. Spend a lot of hours on the same pair so you learn how they translate, and check final mixes on monitors when you can. For dialogue editing, headphones are excellent for catching mouth clicks and small details. More in this video.

Headphones aren't flat, and they shouldn't be. Target curves like the Harman curve try to replicate the sound of good speakers in a good room. Look for:

• Clarity: frequencies shouldn't smear together.
• Impedance your headphone amp can drive: lower impedance is better for laptops and field recorders.
• Comfort: you'll wear them for long sessions.

I prefer open-back headphones for mixing because they feel more like monitors. Use closed-back when recording, so your headphone audio doesn't bleed into the mic. Also look for reviewers who mix dialogue, not audiophile reviewers. More in this video.

It depends on the location, but I find that a soft light source on the "far side" of the person's face along with some sort of accent lighting usually works well. Here's an example video showing how to bounce light off of a piece of fabric to soften it.

Soft light creates smoother transitions between light and shadow, especially on faces. Soft light is created by using LARGE light emitting surfaces like soft boxes or light bounced off of walls or large reflectors or fabric/muslin.

SSI (Spectral Similarity Index), developed by the Academy of Motion Picture Arts and Sciences, measures how closely a light's spectrum matches a reference like daylight or tungsten. I find it more useful than CRI, which uses relatively few color samples. Scores in the 80s or 90s mean you're unlikely to see color rendering errors, while many LEDs sit in the 70s. It matters most for client work where color has to be right, like products and skin tones. If you're happy with your results for casual YouTube videos, you don't need to upgrade. More in this video.

If you mostly light people, bi-color gives you more output for the money, because a full-color light's color engine trades some raw output for color flexibility. Choose RGB if you also want the light for colored accents, background washes or effects. A common setup is a bi-color key light plus a smaller RGB light for the background. More in this comparison.

Try bouncing your light off, or shooting it through, unbleached muslin instead of a bright white surface. Unbleached muslin is a slightly tan, off-white fabric. In my comparison, it gave warmer highlights and less of a magenta cast than a white bounce. You can buy muslin made for lighting, or pick up unbleached muslin fabric at a fabric store. More in this video.

No. I don't accept payment for coverage, and brands don't get to review or approve a video before it's posted. Some products are sent to me free or on loan, and I disclose that in every video; other gear I buy with my own money. I've also examined this publicly, including a 10-week experiment accepting no free gear at all. More in Is Curtis Judd a Shill? and No Free Gear for 10 Weeks.