The Best Microphone for Voice Acting (And What You Don't Need)
The microphone is the cheapest important decision you will make in this business, and it is the one beginners agonise over the longest. Working actors book jobs on mics that cost a hundred dollars, and lose jobs in rooms that would cost a hundred dollars to fix. This chapter is what to buy, what the specs actually mean, and what an engineer is really rejecting your take for.
Every list of the best microphones for voice acting is written by someone who gets paid when you click the link. That is not a conspiracy, it is just how that page makes money, and it produces a very specific distortion: the article is always about which mic, never about whether the mic is the problem. It almost never is.
Here is the shape of the truth. A $119 microphone in a quiet, treated room will pass professional quality control. A $3,750 microphone in an untreated spare bedroom will not. The mic is the cheap part. The room is the hard part. You can spend the next hour reading about capsule diameter, or you can spend it hanging blankets, and only one of those will change whether you book.
So read this chapter as a shopping list with a warning label attached. Buy the mic. Buy it once. Then go and read the next chapter, because that is where your audio problems actually live.
What working voice actors actually use
There are two honest tiers, and the gap between them is smaller than the marketing wants you to believe.
The good-enough-to-book tier. The Audio-Technica AT2020 at $119 is the one I would hand a beginner without hesitation. Around it sit the Shure SM58 at $109, the Audio-Technica AT2040 at $109, the AT2035 at $149 to $159, the Rode NT1 (5th generation) at $214 to $228 with a shockmount and pop filter in the box, and the Lewitt LCT 440 Pure at $299. Every one of these is capable of delivering a file that a client accepts and pays for. None of them will be the reason you lose a job.
The working-pro tier. The Audio-Technica AT4040 at $329. The Shure SM7B at $439. The Electro-Voice RE20 at $449. The Audio-Technica AT4053b at $769. Then the two names you will hear repeated forever: the Sennheiser MKH 416 at $999, which Sennheiser's own newsroom calls the industry standard for boom and voice over, and the Neumann TLM 103 at $1,295. Above those, the Sennheiser MKH 8060 at $1,699 and the Neumann U87 Ai at $3,750, which is the studio standard largely because every professional studio already owns one, so you sound the same everywhere you go.
The 416 against the TLM 103 is the oldest argument in voice over, and the rough consensus is this. The 416 has a lift in the upper mids baked in, which cuts through a mix and is genuinely more forgiving of an imperfect room. It lives in commercial, promo, corporate and ADR. The TLM 103 is warmer and lusher underneath, which suits animation, character and audiobook work, and it turns up wherever someone wanted a U87 and would not spend the money. Mike DelGaudio, a narrator who works for the New York Times, Atlas Obscura and Audible, reportedly owns and uses both. The top of the profession has not settled this argument. You are not obliged to settle it either, at the start of your career, with money you have not yet earned.
Buy XLR, not USB, and the reason is not sound quality
A USB microphone plugs straight into your computer. An XLR microphone uses a three-pin cable and needs a separate box to connect to your computer. The USB one is more convenient and cheaper. Buy the XLR one anyway.
The reason is structural, not sonic. Voice coach Lili Wexu puts it plainly: as you move up, you will be asked to record sessions via Source-Connect, which is the software a commercial studio uses to direct you live, down the line, from their room into yours. Those studios run analog XLR microphones and are not accustomed to USB setups. Nobody rejects your USB mic because it sounds bad. They reject it because when a real studio wants to direct you in real time, your rig does not speak their language. The upgrade costs about $30 more at the mic. Make it now, not the week you finally get the call.
Audio-Technica AT2020 (XLR) — $119. Focusrite Scarlett Solo 4th gen — around $130. An XLR cable and a stand — $25 to $35. Six to eight moving blankets and a PVC frame — $100 to $150. Total: roughly $375 to $435. In a genuinely quiet room, that setup clears the noise-floor and volume specs that professional audiobook QC demands. If you find another $200 down the back of the sofa, spend every cent of it on more and thicker blankets — not on a better microphone. The room is what fails quality control. The mic almost never is.
What an audio interface actually does, in plain words
Your computer has no XLR socket, and even if you could jam the cable in, it could not do the three jobs that need doing. An audio interface is the box that does them. First, it contains a preamp, which takes the very weak electrical signal a microphone produces and boosts it up to a usable level. Second, it contains an analog-to-digital converter, which turns that electrical wave into the numbers your recording software understands. Third, it supplies 48V phantom power, which is the small voltage that condenser microphones need down the cable in order to function at all. If you buy an XLR mic, you need an interface. This is not an upsell. It is the other half of the microphone.
The Focusrite Scarlett Solo (4th gen) is the default answer at $129.99 on sale, $159.99 list. Beyond that: the UA Volt 2 at $179 to $199, the Motu M2 at $199.95, the Focusrite Scarlett 2i2 (4th gen) at $224.99, the SSL 2 MkII at $229.99, the Audient iD4 mkII at $269.99 on sale from $299.99, and the Audient iD14 mkII at $295.15. Any of them will do the job. This is not a place where spending more buys you a booking.
One genuine exception, because it catches people out. The SM7B and the RE20 are dynamic microphones with very low output. To get a usable level from a budget interface you have to crank the preamp near maximum, and a preamp running flat out amplifies its own hiss along with your voice. That is why the Cloudlifter CL-1 at $129 (which adds 20 to 25dB of clean gain) and the TritonAudio FetHead at $79.95 (which adds 27dB) exist. They sit inline between the mic and the interface and let the preamp run cool and quiet. If you buy an SM7B, budget for one of these. It is not an accessory. It is part of the microphone.
The specs a client will actually demand
Start with the term that will be thrown at you first. Your noise floor is the sound of your room when you are not speaking — the fridge two rooms away, the computer fan, the traffic, the faint electrical hum of the building, all of it added together. It is measured in negative decibels, and the more negative the number, the quieter the room. An untreated bedroom typically sits somewhere around -40 to -50 dBFS. A decent home studio baseline is -55dB or better. Professional and broadcast work wants -65dB or lower. Those last two are practitioner consensus rather than a codified rule, but the one that is codified is ACX and Audible's: your noise floor must be no higher than -60dB RMS, and background noise that exceeds it is the single most common reason home recordings fail.
Here is the piece of arithmetic that ruins people. Whatever gain you add in post to bring a quiet recording up to the required loudness, you add to the noise floor as well. If you need 10dB of gain to reach your target, your raw noise floor has to be below -70 dBFS to survive it. If you are sitting at -65 dBFS, you are already five decibels short before you have started. You cannot mix your way out of this. You can only be quieter in the first place.
ACX / Audible: RMS volume between -23dB and -18dB. Noise floor no higher than -60dB RMS. Peaks no higher than -3dB. Files as 192kbps or better CBR MP3 at 44.1kHz. Mono or stereo — despite what half the internet says, stereo is allowed, it simply must be consistent across every file. 1 to 5 seconds of room tone at the head and tail of each file. Maximum 120 minutes per file. And as of 2026, human narration is required: unauthorised text-to-speech or AI recordings are prohibited.
Broadcast and commercial delivery: 48kHz / 24-bit WAV is the standard, with 44.1kHz / 16-bit as the legacy floor. Sample-rate mismatch is one of the most common reasons files get bounced at ingest. Mono is standard for voice over delivery, and you deliver dry — no EQ, no compression, no noise reduction, no gate. The receiving engineer applies their own chain, and a file that arrives pre-mastered marks you instantly as an amateur.
A note of honesty on that -3dB peak figure: it is real and it is everywhere, but it is a marketplace convention — an explicit Voice123 rule, so that files survive their WAV-to-MP3 conversion — and not a broadcast standard handed down by any standards body. Meet it, because clients ask for it. Just do not let anyone tell you it came from the ATSC.
What a take actually gets rejected for
This is the part nobody writes, so here it is. Noise floor above threshold is number one, and it is not close. After that: plosives, those thumps of air on hard P and B sounds. Mic pops. The click of your mouse. Mouth noise. An outtake you forgot to cut. HVAC and computer-fan hum — and the strongest fix there is not software, it is moving the microphone closer to your voice and further from the fan. A 60Hz electrical buzz. Room reverb, the boxy sound. Levels that drift between takes or between chapters.
And then the one that nobody expects: over-processing. More people fail audiobook quality control for cleaning their audio too hard than for not cleaning it enough. A beginner hears noise, panics, and runs noise reduction until the voice sounds like it is underwater. The human doing QC hears those artefacts instantly, and the published guidance is explicit that heavy processing is not liked. Be as gentle as you can, with as few corrections as possible. Fix the room, not the file.
The reflection filter. That curved foam-and-metal shield that clamps behind your mic — the sE Reflexion Filter Pro runs $199.95 to $209. The voice over training company Gravy For The Brain is scathing about it: they say the reflection filter essentially does nothing, that they tested it by recording with and without one and stood aghast at the lack of difference, and that these things are, in their words, a waste of time and money. In fairness, there is credible pushback — commenters on that same piece argue it does reduce the energy hitting the wall behind you. So call it contested. But I would put that $200 into blankets and absorption without a second thought, and so, on the record, would they.
A USB microphone, for the reasons above. A better microphone as a fix for a bad room — the most expensive mistake on this list, because it works in reverse: a good mic in a bad room hears the bad room more clearly. And a closet full of hanging clothes as your only treatment: it helps, a little, but only the very highest frequencies get absorbed by fabric. It does not solve the room.
So: buy the AT2020 and the Scarlett Solo, or something close to them, and stop thinking about microphones. You are not being held back by a capsule. If you want to know what is actually standing between you and a file that gets accepted, it is the six feet of air around your head — and that is the next chapter.
Want Will to Coach You Through It?
Reading is one thing. Working 1-on-1 with a working actor who booked Oppenheimer is another.