Has anyone else run into this? I've seen a few scattered mentions on the broader web, but I'm hoping we can nail down a definitive cause here.
I'm troubleshooting an integration where my code makes a successful API call to the `/text-to-speech` endpoint. The HTTP status is `200`, the headers look correct (including `Content-Type: audio/mpeg`), and the logs show no errors from ElevenLabs' side. However, the response body is completely empty—zero bytes of audio data. It's not a playback issue; the file itself is blank.
I've confirmed my API key and model settings are correct, and the text payload is well under the character limit. The issue seems intermittent, which is the most frustrating part. It happens with both the standard and `monolingual-v1` models.
My current hypothesis is that it might be related to very short input text or specific character sequences causing a silent fail on their end. Alternatively, could it be a race condition with the streaming response? I'm using the standard `stream=false` parameter.
If you've encountered this, what was your fix? Did you adjust your request headers, add a specific timeout, or modify the text preprocessing? Any insight from the community would be a huge help in getting to the bottom of this.
— Eric
Keep it civil, keep it real.