How to eq your podcast voice in the browser
- Step 1Sign in on a Pro plan — Voice EQ is a Pro-tier tool (
minTier: pro). On the Free plan the audio family is capped at 50 MB / 30 minutes and this tool is locked, so confirm you're on Pro before you start. Pro raises the per-file limit to 200 MB and the per-file duration limit to 120 minutes. - Step 2Drop your podcast voice file onto the tool — Add one audio file — the tool processes a single file per job (no batch). MP3, WAV, FLAC, M4A, OGG, and Opus all decode. Best results come from a clean, single-voice track; for a multi-track episode, EQ each host's track separately before you mix down.
- Step 3Pick the Voice profile that matches your host — Choose Male (high-pass 80 Hz, +3 dB at 3 kHz), Female (high-pass 100 Hz, +3 dB at 4 kHz), Bright (broadcast) (high-pass 80 Hz, +4 dB at 5 kHz), or Warm (high-pass 70 Hz, +2 dB at 2.5 kHz). There are no manual sliders — each profile is a fixed curve, so try one, listen, and switch if it's not right.
- Step 4Process the file — FFmpeg runs
highpassplus two peakingequalizerbands as a single-affilter chain in your browser. A typical 30-60 minute episode finishes in well under a minute on a modern laptop; the very first run is a few seconds slower while the WASM engine loads. - Step 5Download and A/B against the original — Download the EQ'd file and compare it to the source before you publish. EQ is subjective — if the Male profile sounds thin on your voice, try Warm; if you still sound muddy, the room (not the EQ) is the limit.
- Step 6Slot EQ into the right point of your chain — Best order: denoise first with the ai-noise-reducer so EQ isn't boosting hiss, then EQ here, then level with the speech-leveler and finish on the loudness-normalizer for a -16 LUFS podcast target. The all-in-one podcast-master runs that whole chain at once.
The four Voice profiles — exact FFmpeg curves
Every profile is a fixed three-stage chain: one high-pass plus two peaking equalizer bands (each Q=1, the FFmpeg default width for t=q:w=1). Values are read directly from the processor — there are no user-editable frequency or gain controls.
| Profile | High-pass | Low-mid cut | Presence/air boost | Best for |
|---|---|---|---|---|
| Male (default) | highpass=f=80 | -2 dB @ 300 Hz | +3 dB @ 3 kHz | Most spoken male voices; the safe starting point |
| Female | highpass=f=100 | -2 dB @ 300 Hz | +3 dB @ 4 kHz | Higher fundamentals — higher HPF and higher presence band |
| Bright (broadcast) | highpass=f=80 | -1.5 dB @ 250 Hz | +4 dB @ 5 kHz | A forward, radio-style voice; the most aggressive presence lift |
| Warm | highpass=f=70 | -1 dB @ 350 Hz | +2 dB @ 2.5 kHz | Thin or harsh voices — lowest HPF keeps more body, gentlest curve |
What podcast problem each stage of the chain solves
The same three-stage architecture in every profile, mapped to the studio problem it fixes. The whole chain applies in one pass.
| Stage | FFmpeg filter | Podcast problem it solves |
|---|---|---|
| High-pass | highpass=f=70..100 | Removes desk thump, HVAC rumble, traffic, and mic-handling noise below the voice — a 2nd-order (12 dB/oct) roll-off |
| Low-mid cut | equalizer=f=250..350:t=q:w=1:g=-1..-2 | Clears the boxy, congested 'recorded in a small room' build-up around 300 Hz |
| Presence boost | equalizer=f=2500..5000:t=q:w=1:g=2..4 | Adds clarity and intelligibility so consonants survive on phone speakers and earbuds |
Tier limits for Voice EQ (audio family)
Voice EQ requires Pro. Limits are per file; the tool is single-file (no batch). Duration limit is separate from the size limit.
| Plan | Max file size | Max duration | Voice EQ available? |
|---|---|---|---|
| Free | 50 MB | 30 min | No — Pro feature |
| Pro | 200 MB | 120 min | Yes |
| Pro-media | 100 GB | Unlimited | Yes |
| Developer | 100 GB | Unlimited | Yes |
Cookbook
Profile choices and chains for real podcast scenarios. Each shows the exact FFmpeg filter the tool runs so you can see precisely what changed.
Solo male host, untreated home office
Default Male profile. The 80 Hz high-pass kills the desk and chair rumble; the -2 dB at 300 Hz clears the small-room box; the +3 dB at 3 kHz brings the voice forward.
Profile: Male (default) FFmpeg chain (single -af pass): highpass=f=80, equalizer=f=300:t=q:w=1:g=-2, equalizer=f=3000:t=q:w=1:g=3 Input: episode-host.wav (WAV in) Output: episode-host-eq.wav (WAV out, re-encoded PCM s16le)
Female co-host who sounds buried in the mix
Female profile. The high-pass moves up to 100 Hz to match the higher fundamental, and the presence boost moves up to 4 kHz so the voice cuts through without sounding harsh.
Profile: Female FFmpeg chain: highpass=f=100, equalizer=f=300:t=q:w=1:g=-2, equalizer=f=4000:t=q:w=1:g=3 Input: cohost.mp3 -> Output: cohost-eq.mp3 Note: MP3 in is decoded and re-encoded (libmp3lame) — one extra lossy generation. Keep a WAV master if you have one.
Radio-style intro VO that needs to be 'in your face'
Bright (broadcast) profile — the most forward curve, with the biggest presence lift (+4 dB) pushed up to 5 kHz for a polished, announcer sound.
Profile: Bright (broadcast) FFmpeg chain: highpass=f=80, equalizer=f=250:t=q:w=1:g=-1.5, equalizer=f=5000:t=q:w=1:g=4 Use it on the cold-open VO; consider Male/Female for the main conversation so the whole episode isn't this bright.
Thin, slightly harsh lapel-mic voice
Warm profile. It uses the lowest high-pass (70 Hz) so more of the voice's body survives, the gentlest low-mid cut (-1 dB), and a lower, softer presence boost (+2 dB at 2.5 kHz) to avoid making harshness worse. Note: warmth here comes from keeping low end, not from adding a bass boost band.
Profile: Warm FFmpeg chain: highpass=f=70, equalizer=f=350:t=q:w=1:g=-1, equalizer=f=2500:t=q:w=1:g=2 There is NO low-shelf bass-boost stage. 'Warm' = less is cut and less is removed, not 'bass added'.
Right place in the publishing chain
EQ is one link. Run denoise first so EQ doesn't amplify hiss, EQ second, then level and normalise. The podcast-master tool chains all of this in one job.
Recommended episode chain: 1. ai-noise-reducer (RNNoise — remove room hiss/hum) 2. voice-eq (THIS TOOL — tonal balance) 3. speech-leveler (even out level word-to-word) 4. loudness-normalizer (EBU R128, target -16 LUFS) Or run all four at once with /audio-tools/podcast-master
Edge cases and what actually happens
No frequency or gain sliders appear
By designVoice EQ exposes exactly one control: the Voice profile picker (Male, Female, Bright, Warm). Each profile is a fixed curve baked into the processor — there is no manual band, frequency, Q, or gain input. If you need a fully parametric, draw-your-own EQ, this tool isn't it; it's a curated chain for spoken voice.
Free-plan user can't open the tool
Pro requiredDespite 'free' appearing in the URL for SEO, Voice EQ is gated at minTier: pro. On the Free plan the tool is locked and the audio family caps at 50 MB / 30 minutes. Upgrade to Pro (200 MB / 120 min) to run it.
MP3 in, MP3 out loses a little quality
ExpectedEQ requires re-encoding. The output container matches the input, so an MP3 source is decoded, EQ'd, and re-encoded with libmp3lame — one extra lossy generation. Inaudible for most spoken content, but if you have a WAV or FLAC master, EQ that instead and export the lossy version last.
Episode is longer than your plan's duration limit
400 — duration exceededThe audio family enforces a per-file duration cap separate from size: 30 min on Free, 120 min on Pro, unlimited on Pro-media/Developer. A two-hour interview on Pro is over the 120-minute limit even if it's under 200 MB — split it with the audio-splitter, EQ each part, then rejoin with the audio-merger.
File is larger than your plan's size limit
413 — too largePro caps at 200 MB per file. A long WAV master can exceed that quickly. Convert to a high-bitrate compressed format first, or move to Pro-media (100 GB) for full uncompressed masters.
EQ makes the boxiness slightly better but it's still muddy
Room limitThe -1 to -2 dB cut at 250-350 Hz is gentle by design. Heavy room reverb or a very boomy space can't be EQ'd away — fix the recording with acoustic treatment, mic placement, or denoise. EQ shapes tone; it doesn't remove reflections.
Picked Warm expecting added bass
By designThe Warm profile has no bass-boost band. It sounds warmer only because it cuts the least (-1 dB) and uses the lowest high-pass (70 Hz), preserving more low end than the other profiles. If you want genuine low-end body added, this chain won't do it — there's no low-shelf boost in any profile.
Multi-track episode EQ'd as one mixed file
SuboptimalThe profiles assume a single voice. If you EQ a mixed-down episode with two very different voices, the curve fits one better than the other. EQ each host's isolated track separately before mixdown for the cleanest result.
Stereo music bed in the file
SupportedThe filters apply across all channels, so a stereo file stays stereo and any music bed gets the same EQ. The chain is voiced for speech, so a heavy presence boost may make a music bed sound thin — EQ the voice before the bed is added if you can.
Corrupt or unsupported file
Decode errorIf FFmpeg can't decode the input (truncated download, mislabelled extension, DRM-protected file), the job fails before EQ runs. Re-export a clean MP3/WAV/FLAC from the source and try again.
Frequently asked questions
What EQ does this apply for a podcast voice?
A three-stage chain: a high-pass to remove rumble, a gentle peaking cut around 300 Hz to clear boxiness, and a peaking presence boost around 3-5 kHz for clarity. The exact frequencies and gains depend on the profile you pick — Male, Female, Bright, or Warm.
Can I set my own frequencies and gain?
No. There's a single Voice profile picker with four fixed curves. It's a curated chain for spoken voice, not a parametric EQ. For draw-your-own EQ you'd need a full DAW.
Is it really free?
It's free of uploads and cloud cost — everything runs in your browser — but the tool itself requires a Pro plan. The Free tier doesn't include Voice EQ.
Does my audio get uploaded?
No. Processing runs entirely in your browser on FFmpeg 8.1 compiled to WebAssembly. The file never leaves your device, which is ideal for unreleased or embargoed episodes.
Which profile should I start with?
Male is the default and the safest starting point for most spoken voices. Use Female for higher voices, Bright for a forward broadcast sound, and Warm for thin or harsh recordings.
Should I EQ before or after noise reduction?
Denoise first, then EQ. Denoising first stops the presence boost from amplifying hiss. Run the ai-noise-reducer, then this tool, then leveling and loudness normalisation.
Will EQ fix a bad microphone or a bad room?
Partially. The high-pass helps with rumble and the low-mid cut helps with boxiness, but heavy reverb or a fundamentally dull mic can only be improved so far. Fixing the source always beats fixing it in EQ.
What formats can I drop in?
MP3, WAV, FLAC, M4A, OGG, and Opus all decode. The output uses the same container as the input.
Does it change the file format?
No — output matches input. MP3 in gives MP3 out (re-encoded with libmp3lame), WAV gives WAV (PCM 16-bit), FLAC gives FLAC. To convert formats, use a sibling converter like wav-to-mp3 first or after.
How long can my episode be?
Up to 30 minutes on Free (where the tool is locked anyway), 120 minutes on Pro, and unlimited on Pro-media and Developer. The duration cap is separate from the 200 MB Pro size cap.
Can I EQ several episodes at once?
No — Voice EQ processes one file per job. Run them one at a time, or use the multi-file podcast-master if you want a full chain across several episodes.
What's the difference between this and a full podcast master?
This tool only does tonal EQ. The podcast-master runs denoise, silence stripping, EBU R128 loudness normalisation, and a true-peak limiter as well — it's the one-click 'publish-ready' option.
Privacy first
Every JAD Audio tool runs entirely in your browser via FFmpeg (WebAssembly) and RNNoise. Your audio files never leave your device — verified by zero outbound network requests during processing.