A pre-release version for version 1.8.0 has been made available. Please keep in mind that this is not the final version of 1.8.0 and may include bugs or other issues.
Pre-release versions are available for download from the Dreamtonics website, and cannot be installed from within the application. The same applies for compatible voice databases (previous voice versions will still work the same).
To summarize the changes:
- Diffusion Probabilistic Models (DPM) offer improved realism, especially in vibratos
- Notes can now be entered with a MIDI keyboard
- There is now a built-in metronome function
- Note-level cross-lingual synthesis allows seamless transition between languages without the need for a new group or track
- the voice database Mai is available for free for all Pro edition users (Synthesizer V AI Mai: New Feminine Japanese Voice from Dreamtonics - shiwei_migi の #4)
Thanks Claire… Look forward to seeing responses to the beta. I will not upgrade yet as I have several tracks underway which I cannot afford to risk. Thanks again!
Here are the introduction videos to go along with tbe announcement. The YouTube and bilibili videos are a little different (For YT, Yuma was used to demo DPM and Mai sings in English for note-level cross-lingual, while for BB, it’s An Xiao for DPM and Mai singing in Mandarin Chinese)
Is there a way to test out the DPM now? I believe that only the latest beta versions of voicebanks can use it, so I’m wondering if there’s a beta of Yuma, so that this feature can be used.
Yes, the beta version installers are available here (including Yuma):
I absolutely LOVE that they’re giving us a free voicebank. Normally I don’t like messing with beta software because I’m afraid I’ll wreck my existing setup, but I had to hear the new update along with the new voicebank. I’ve played around a bit with the beta update of SOLARIA and Qing Su–the DPM update improvement is noticeable.
Hi… Thanks for the feedback… I was worried that my Solaria tracks would sound different after upgrading. I gather from your comment that there are no negative side effects? What about instances where you have manually edited notes/curves etc? Does the new version honor these manual edits or does it re-render everything from scratch?
It shouldn’t modify your SVP, though the rendered result may sound different since the updated engine interperets your project slightly differently. Reactions I’ve seen on social media have been generally positive so far.
All that said, it is a pre-release version so it would be wise to back up your SVP before using it in 1.8.0b1, even if just to be completely safe from potential bugs while you see how things have changed.
edit: the one exception I just thought of is if Instant Mode is enabled. In this case I expect it would recompute the auto-pitch tuning once a change is made.
Excellent insight. Thanks Claire!