Before you reach out: a few things to try
Most problems come down to a handful of common causes, so it's worth running through these before writing in:
- Try a WAV version of your file. Some issues trace back to a specific codec or container rather than the tool itself.
- Try a shorter clip. If a long file times out or runs out of memory, a shorter test tells you whether it's a length or memory issue.
- For AI tools — voice cloning, stem splitter, text to speech, speech to text, and AI-mode vocal remover — use a desktop build of Chrome or Edge. These give the most consistent WebGPU support for the heavier models.
- Keep the tab open while a tool is processing or downloading a model. Closing or reloading it mid-task interrupts both.
- If a tool that worked before starts behaving oddly, clear its cached model through Studio settings and let it download fresh.
What to include in a report
If those steps don't fix it, a clear report is the fastest way for us to reproduce and fix the problem. Please include:
- The exact tool name and what you were trying to do.
- Your browser and its version, and your operating system.
- Your file's format and rough length (for example, "MP3, about 4 minutes").
- The exact error text or a description of what went wrong, including at what step it happened.
One thing to leave out
Please don't send us your audio file, especially not a private recording. Since everything processes on your own device, we can rarely do anything useful with the file itself, and describing the problem in text is both enough to diagnose it and safer for you.
Get in touch
For support, bug reports, or general questions, write to the contact address published on this page. If you're reporting a privacy or security issue specifically, send it to the same address with "Privacy" or "Security" at the start of the subject line, so it gets triaged correctly.