Uneven volume between speakers in an interview
Checked against the product on October 22, 2026
When one speaker is much louder than the other, viewers keep reaching for the volume and captions suffer for the quiet voice. The causes are microphone distance, different input levels and a shared microphone. Balance levels in editing software, avoid boosting noise and record each person on their own microphone with matched levels next time.
What you are seeing
The host booms and the guest whispers, or the reverse. In a clip, the viewer adjusts the volume at every turn. The quiet voice may also be transcribed less accurately, so captions contain more errors there.
Common causes
- Different distances from the microphone.
- One input level set higher than the other.
- A shared microphone placed nearer one person.
- A remote guest with a weak connection or microphone.
- Different speaking styles, soft or loud.
How to fix it
- Listen to both voices separatelyFind the gap.
- Raise the quiet speaker's levelWithin reason.
- Lower the loud speaker slightlyAvoid clipping.
- Check noise on the quiet voiceBoosting can raise hiss.
- Review captions on the quiet partsCorrect errors.
- Check on a phone speakerDoes the balance hold?
Do not over-boost
Raising a very quiet recording also raises its noise. If the voice is too faint, consider whether another passage would make a better clip.
How to avoid it
Give each person a microphone at a similar distance, set matching levels in a test and ask remote guests to use a headset. A minute of level-matching saves much of the editing.
In Averoai.pro
Averoai.pro builds captions from the audio you provide, and clearer, more even speech leads to fewer word errors. It does not mix audio, so balance levels in your recording or audio software and then review the captions of the quieter speaker.
A worked example
A podcast guest recorded on a laptop microphone while the host used a headset. After evening the levels and trimming a noisy intro, the clip played comfortably, and the captions for the guest needed far fewer corrections.
Worth remembering
Even levels are a kindness to viewers and to transcription. Match microphones, check levels before you record and avoid boosting noise. When one voice remains too faint, choose a different moment. Make a habit of recording ten seconds of each person before the interview starts, listening on headphones and adjusting the levels there and then, because fixing the balance at the source takes seconds while repairing it afterwards can take an hour and never sounds quite as natural.
Frequently asked questions
Why is one speaker quieter?
Different distances, input levels or microphones.
Can I fix it in editing?
Partly; avoid boosting noise.
Does it affect captions?
The quiet voice may be transcribed less accurately.
How do I prevent it?
Give each speaker a microphone and match levels in a test.
Features that help with this
Related reading
- Caption errors with strong accents and regional speech
Automatic transcription can stumble on accents, dialect words and fast speech. Learn how to set the language, improve audio, proofread and keep a word list.
- A talking-head clip that feels visually static
One unchanging shot can tire viewers even when the words are good. Add light variety with framing, cutaways and captions without gimmicks or false visuals.
- What is an A/B test?
An A/B test compares two versions that differ in one thing. Learn how to run a fair test with clips, why small samples mislead and what to conclude.
- Too many filler words like um and uh in a clip
Filler words and false starts slow a clip and distract viewers. Learn how to trim them, keep speech natural and avoid choppy edits in your captions.
- Caption styles that differ from clip to clip
A series looks messy when fonts, colours and positions change from clip to clip. Learn how to define a style once, apply it consistently and check each export.
- Audio that distorts on loud laughs and shouts
Crackling or harsh sound on loud moments comes from levels hitting the ceiling. Learn how to lower gain, recover what you can and record safely next time.
Also available in Français