Audio ad specs: Spotify, Amazon, SiriusXM, and YouTube

Audio ad specs are set by each seller. Spotify, Amazon, and SiriusXM Media (whose ad portfolio lists Pandora) all run spots of 30 seconds or less as standard, but they differ on files: Spotify wants MP3, WAV or OGG at 44.1 kHz, 192–320 kbps, and -16 LUFS; Amazon wants WAV, MP3 or OGG of 10–30 seconds and up to 3 MB, normalized to -14 dBFS RMS; SiriusXM wants an MP3 at 44.1 kHz. A YouTube audio ad is uploaded as a video, not an audio file.

Each seller's rules are quoted from its own page, read on 2026-09-29: Spotify, Amazon Ads, SiriusXM Media, and Google's YouTube audio line items page. Sellers can change them. Sume facts come from the Audio detach and Timeline 1.0 docs and the Sume API reference.

How do audio ad specs differ by seller?

A dash means the seller's page does not state that spec.

How long should an audio ad be?

Thirty seconds fits every seller above. The exceptions, plus one script rule:

• Spotify: long-form music ads of up to 60 seconds exist in select markets for Direct campaigns. Spotify recommends no more than 65 words for 30 seconds and 100 words for 60.
• Amazon: any length from 10 to 30 seconds, but an ad meant for Twitch inventory must run :12–:15 or :27–:30.
• SiriusXM: 45-second ads are for pharma brands, and 45 and 60 seconds are available for programmatic pharma campaigns on Pandora mobile audio only.
• Amazon also requires that ads not mention click-through actions, for compatibility with smart speakers.

Can one audio file meet every seller's specs?

No. Make one mix, then export per seller. A 30-second, 44.1 kHz, 16-bit stereo WAV is about 5.29 MB, over Amazon's 3 MB cap, and its bitrate is above Spotify's 320 kbps maximum. A 30-second MP3 at 192 kbps is about 0.72 MB and sits inside both Spotify's and Amazon's bitrate rules. The level targets differ too: Spotify measures loudness in LUFS, while Amazon states RMS and peak levels in dBFS, so each seller's master needs its own level check. YouTube needs a video file, not audio; YouTube audio ads covers that upload.

How do I make an audio ad with Sume?

Voice the script with text to speech, mix a music bed under it in a Timeline 1.0 render, and take the sound out with audio detach; AI radio commercial generator walks through the jobs. What each step can and cannot set:

• Text to speech: output_format can ask for mp3 at sample_rate 44100 and bit_rate 192000, the highest MP3 rate the schema lists.
• Audio detach: wav (pcm_s16le) or mp3 at 128 kbps, which is under the 192 kbps that Spotify and Amazon ask for. sample_rate can be 44100, and channels is source or mono.
• Levels: Timeline has audio.gain_db (−60 to 12) and soundtrack gain_db and duck_db, but the docs list no loudness target, so measure LUFS and dBFS and make each seller's final encode in an audio tool.