Looking at your workflow and the Hume AI API documentation, the issue is likely with how the binary data is being handled after the HTTP request. Based on the [dev.hume.ai]( Text-to-Speech (File) | Hume API ) documentation, Hume returns audio files in the specified format (MP3 in your case).
Try these steps to fix the binary data handling:
• In your HTTP Request node, make sure “Response Format” is set to “File” (not JSON) since you’re expecting an audio file
• Check that the “Binary Property Name” in the HTTP Request node is set to “data” (the default)
• In both Box and Dropbox nodes, ensure the “Input Data Field Name” matches the binary property name from the HTTP Request (usually “data”)
• If you’re still having issues, add a Code node between the HTTP Request and upload nodes to verify the binary data structure with `console.log(Object.keys($input.all()[0].binary))`
The [community.n8n.io]( How to use binary data in Http Request ) discussions show this is a common issue with binary file handling - the key is ensuring the binary property names match between nodes.