Clips Live on Phone Speakers: A Loudness Guide for Short-Form Audio

Most Clips Are Mixed for the Wrong Room

A podcast is mixed for earbuds and a quiet car. A clip from that podcast plays on a phone speaker, in a feed, right after something mastered to be loud. Viewers who reach for the volume are already halfway to the next swipe.

Picture quality gets a checklist. Audio usually gets whatever the episode had. This post is the checklist.

Nobody Publishes a Short-Form Target

We looked for a published loudness target from TikTok, Instagram and YouTube Shorts and did not find one. What does exist are documented references from neighboring media, all measured the same way, under the ITU-R BS.1770 recommendation for measuring programme loudness and true-peak level (ITU-R BS.1770):

• Broadcast. EBU R 128 sets a target of -23.0 LUFS for programme loudness, with a tolerance of plus or minus 1 LU where the target cannot practically be met (EBU R 128).
• Podcasts. Apple recommends that "the overall loudness remains around -16 dB LKFS, with a +/- 1 dB tolerance, and that the true-peak value doesn't exceed -1 dB FS" (Apple Podcasts for Creators).
• Music streaming. Spotify says "we adjust tracks to -14 dB LUFS, according to the ITU 1770 standard," and leaves 1 dB of headroom for lossy encodings (Spotify for Artists).

LUFS and LKFS are the same unit under two names. Integrated loudness is measured across the whole clip, not at its loudest moment, which is why a clip can peak hot and still sound quiet.

Why a Quiet Clip Stays Quiet

Spotify's documentation shows how normalization treats quiet material, and the logic is worth borrowing. It turns loud tracks down freely, but it will not lift a quiet track past its headroom: a track at -20 LUFS with a true peak of -5 dBFS is only raised to -16 LUFS (Spotify for Artists). Whatever a short-form app does internally, the safe assumption is the same. Normalization can save a loud clip from being too loud. It cannot be relied on to rescue a quiet one.

Our working target, in the absence of a published one, is the podcast reference: about -16 LUFS integrated, true peak at or below -1 dBTP. It is loud enough to sit next to other speech in a feed and leaves headroom for the platform's encoder. Treat it as a starting point and check it on a phone.

Speech Over Music

Music under speech is the other common failure. WCAG's guidance for prerecorded audio-only content sets a useful line: background sounds "at least 20 decibels lower than the foreground speech content," because "people who are hard of hearing may have difficulty distinguishing speech from music and other sounds" (W3C, Understanding SC 1.4.7). That criterion is written for audio-only content and sits at Level AAA, so it is not a legal bar for video. As a mixing rule of thumb for a music bed under a talking head, it holds up well.

A Clip Audio Chain in Five Steps

1. Clean first. Remove hum, hiss and room noise before touching loudness, or the loudness step raises the noise with the voice.
2. Level the speakers. On a two-person clip, one quiet guest and one loud host is the most common complaint. Match them before anything else.
3. Compress gently. Enough to keep speech even across a phone speaker, not so much that breaths come up to the level of words.
4. Set music well under speech. Aim for the 20 dB gap, and duck the bed further under any sentence that carries the point.
5. Measure the clip, not the episode. A 40-second excerpt from a normalized episode is not itself at target, because integrated loudness is measured over whatever you measure. Normalize each clip, limit true peak to -1 dBTP, then listen once on a phone speaker.

Where ClipForge Fits

ClipForge cuts clips from the audio you upload, so a clean episode mix is the best input you can give it. The Pro plan adds an audio mixer alongside colour, multicam and VFX (pricing), and the User Guide walks through the editor. Whatever tool does the mixing, measure loudness on the exported clip, not on the source.

Keep Reading

• How to Edit Vertical Video for Maximum Watch Time: The 2026 Retention Playbook
• Podcast-to-Video: How to Turn Audio Episodes into 20+ Short-Form Clips Without Re-Recording
• Captions That Count: What WCAG and the ADA Ask of Short-Form Clips

, Rocky