Troubleshooting · How we test and maintain this guide
The failed check tells you where to look
Preparation does more than attach an audio extension to a response. It reads the transferred bytes, identifies the container, asks the browser to decode them, compares duration with the source metadata and checks for extremely low sample levels. An error at one stage should not be treated as proof of a problem at every later stage.
Save other ready tracks before investigating a failed row. A correct result can coexist with a bad source response. Record the exact message rather than describing every failure as a broken download. In particular, an HTML error page is not a damaged MP3, and changing its filename cannot turn it into music.
Results from deliberately controlled inputs
| Test input | Observed outcome |
|---|---|
| normal-original | Ready; source bytes preserved |
| duration-mismatch | Audio duration does not match the song. The transfer may be incomplete. |
| silent | The file appears to be silent. No download was prepared. |
| unrecognized-container | The downloaded resource is not a recognized audio file. |
| too-short | Audio transfer was incomplete. Try again. |
| html-error | The audio server returned an error page. |
The normal case used a self-generated twelve-second signal. The mismatch case supplied that same signal with metadata claiming 120 seconds. The silent case used two seconds of zero-valued PCM samples. Other cases supplied unrecognised bytes, a 512-byte response, or a declared HTML error page. These tests exercise specific branches; they do not simulate every way a real network transfer can fail.
Respond to the actual failure
Incomplete transfer: check the connection and whether the source remains accessible. The tool rejects a body below its minimum size and a known length that does not match the bytes received. A normal network retry can help a transient interruption. Continuous retries will not repair a source that is refusing access or rate limiting requests.
Unrecognised resource or error page: return to the source song and confirm the link and its availability. A response can be an error document even when a URL looks like media. Do not feed credentials or an arbitrary replacement media host into the tool. If the same public song repeatedly produces this result, report the message and stage to support.
Cannot decode: the browser may not support that particular audio encoding, or the data may be damaged. The tool needs to decode audio for its checks even in Original mode. Selecting Original therefore does not guarantee a bypass of the browser's decoder. Use an available official export or a compatible current browser for a permitted source when appropriate.
Duration mismatch: the current rule compares decoded duration with the song metadata and rejects a difference above the larger of three seconds or three percent. For a 120-second metadata duration, that threshold is 3.6 seconds. This rule catches a substantial discrepancy; it does not identify its cause or prove that every file inside the tolerance is artistically complete.
Appears silent: the current scan samples channel values at regular intervals and rejects a very low peak. The controlled all-zero signal triggered that check. This is a technical guard, not a loudness measurement or a substitute for listening. A very quiet intentional recording and an unexpectedly silent result need context from the source and your own playback check.
Passing the checks still requires a listening check
A technically valid file can be the wrong take, have an unwanted ending, or contain a musical defect that no container or duration check can judge. Preview the prepared result, then open the saved file in its destination player. Check the beginning, a middle passage and the end before relying on a collection for an editing project.
Our separate format test fully decoded twenty generated outputs with an independent decoder and checked the archived bytes. It establishes those files' technical readability. It does not establish Suno catalogue availability, perceived quality, copyright ownership, or compatibility with every application. Use the measured results to understand the checks rather than as a universal pass certificate.
For a persistent failure, provide the browser version, visible error, whether the source plays normally and whether the problem affects one permitted song or several. Avoid sending private audio or temporary access tokens. A small reproducible case is enough to start an investigation.