Hi vegueire,
Thanks for flagging this and for sharing the files, really helpful for tracking this down.
We took a close look, and the good news is the audio levels themselves are working as expected. What's likely happening is that a prompt like this can produce a result that feels more distant or soft, even when the levels are technically fine, which can come across as "quiet" depending on your speakers or headphones.
To get a stronger, more present result, try being a bit more direct about the strike and resonance in your prompt. A few examples that tend to work well (with results attached below):
- "Single heavy bronze funeral bell toll, deep bass resonance, powerful strike, long ominous decay"
- "One massive church bell toll, strong attack, deep low resonance, dark and powerful"
- "Cinematic funeral bell, single slow heavy toll, deep resonant bass, sharp initial strike"
Also, if you generate a take you like the character of but want it to hit a bit harder, you can raise the gain/volume on that clip in your editor of choice (e.g. Premiere) as a quick workaround.
Hope this helps, let us know how it goes!
Hajar
[Moved to questions]