Audio Level Standards for Video

Loudness normalization replaced the older peak-based approach to audio metering. Instead of measuring the loudest instantaneous peak, modern standards measure perceived loudness over the duration of the program. This approach ensures that a quiet dialogue scene and an action sequence are perceived at similar overall levels by the viewer, eliminating the need to constantly adjust volume.

The unit of measurement is LUFS (Loudness Units Full Scale), sometimes written as LKFS. A typical broadcast target is -24 LUFS (ATSC A/85 for North America) or -23 LUFS (EBU R128 for Europe). Streaming platforms have their own targets: YouTube normalizes to approximately -14 LUFS, Spotify uses -14 LUFS, and Apple Music targets -16 LUFS. Content that does not match the platform target will be adjusted, potentially in ways that compromise the mix.

Three key measurements define loudness compliance: Integrated Loudness (the average over the entire program), Loudness Range (the dynamic spread between quiet and loud sections), and True Peak (the maximum instantaneous sample value, which must stay below a specified ceiling to prevent clipping during encoding). All three must be within specification for the content to pass quality control.

EBU R128 is the European Broadcasting Union standard for loudness normalization. It targets -23 LUFS integrated loudness with a permitted deviation of plus or minus 1 LU. The True Peak maximum is -1 dBTP, providing headroom to prevent inter-sample peaks from causing distortion during codec conversion. The standard also specifies a Loudness Range (LRA) target, though this is advisory rather than mandatory.

ATSC A/85, used primarily in North America, targets -24 LUFS with similar True Peak requirements. While the 1 LU difference between EBU and ATSC may seem trivial, it can cause QC failures when content intended for one market is delivered to the other without adjustment. Always verify which standard applies to your specific deliverable.

Both standards also address dialogue loudness specifically, recognizing that viewers judge overall loudness primarily by how they perceive speech. Dialog gating, where loudness measurement pauses during non-dialogue sections, ensures that the perceived loudness of speech is consistent even in programs with wide dynamic range in their music and effects.

A loudness meter is an essential tool for any video production facility. Hardware and software meters that comply with ITU-R BS.1770 provide accurate LUFS, LRA, and True Peak readings. Many NLE (Non-Linear Editing) applications include built-in loudness meters, but standalone meters often provide more detailed analysis and logging capabilities.

Monitoring environment matters as much as the meter itself. Audio should be evaluated in a treated room with calibrated speakers at the correct reference level. The SMPTE standard for film mixing calibrates each speaker to 85 dB SPL with pink noise at -20 dBFS. For broadcast and smaller rooms, lower reference levels are common, but consistency is the key: always mix and evaluate at the same calibrated level.

Automated loudness analysis can scan an entire program and generate a detailed report showing moment-by-moment loudness, integrated values, and any exceedances. Running this analysis as part of your QC workflow catches issues before delivery. Tools like RecReview can flag audio issues alongside visual ones, providing a comprehensive QC view of the entire deliverable.

Different delivery platforms have different audio requirements beyond loudness. Broadcast may require specific channel configurations: stereo (L/R), 5.1 surround (L, R, C, LFE, Ls, Rs), or Dolby Atmos. Each channel layout has its own loudness and level requirements. Downmix compatibility must also be verified: does the 5.1 mix fold down to stereo correctly without phase cancellation or level imbalance?

Streaming platforms generally accept stereo and may support surround formats. However, their loudness normalization algorithms vary. Content that is delivered louder than the platform target will be turned down, potentially reducing dynamic impact. Content delivered quieter than target will be turned up, potentially raising the noise floor. Delivering at the correct level preserves the intended dynamics.

Social media platforms often play audio on small speakers or headphones, which changes how the mix is perceived. Low-frequency content that sounds powerful on studio monitors may be inaudible on phone speakers. While you should not compromise your mix for the lowest-common-denominator playback device, being aware of how your content will be consumed helps you make better mixing decisions.