editor vs export audio.zip (478.0 KB)
I’ve composed a song and spent time adjusting phenomes, timings, pitch variations and other parameters until I have a pretty good sounding vocal. When this is exported (48k, 32 float or 16bit) the WAV file is not even close to the sound in the editor. It sounds like a poor synth approximation (worse than that other well known Vocal-oid synth).
No matter what settings I use the audio from the render is substandard to the quality of the editor and also a bit quieter.
Anyone else come across this and is there a setting I am missing?
The zip has an example of the type of artifacts compared to the editor version - there are no ASP elements in the exported version (they actually make the export sound a whole lot worse!)
That certainly is odd… Usually the final rendered wav file is the best way to ensure a higher quality output.
- Are you using the software standalone or as a plugin in your DAW?
- Do your render settings (in the Render panel) exactly match the live audio and live rendering settings (in the Settings panel)?
- Have you tried setting the “Render Mode” in the Voice panel to “Prefer Quality”? (this is probably not the problem)
- Have you tried reproducing the issue with a different voice database?
Setting Aspiration Output to “None” is probably best, since that function is only useful if you plan to do different processing on the aspiration and phonation parts separately and then combine them later.
If none of that resolves the issue it would probably be a good idea to reach out to Dreamtonics directly ([email protected]), or perhaps some other forum member has a better ear for what’s going on than I do and can offer some insight.
thanks for the amazing response time again - Dreamtonics should certainly take a leaf out of your book (not even had an acknowledgement to the query I raised last week!)
1 - I can’t use as VST plug in (currently querying with Dreamtonics directly) so used as standalone ATM.
2 - ASIO driver, Settings and Render all set to 48k /1024 buffer
3 - Render on voice panel is set to ‘prefer quality’
4 - I have Kevin, Solaria lite and Solaria full - all have the same issue on export.
I had the ASP set to ‘separate files’ but I will try again with it set to none to see if that makes any difference.
I ended up using loop-back to capture the audio from the editor using Audacity then saved as broadcast wave and then imported them into my DAW. Its an awful long winded process that shouldn’t need to happen but at least I got a reasonable result (have a listen to the short extract which i feel is pretty close to a real human voice )
do you remember cut down example.zip (481.5 KB)
I just tested exporting a test track (Isolated Asp. as seperate track). Did you put both tracks together when mixing? If you don’t put the ASP track with the base then it will sound really bad.
There are benefits to having them separate for editing but when in doubt just render it as 1 together.
The isolated ASP file was in its own track as per the instruction video on Dreamtonics site and mixed low. I’ll give your suggestion a go 'cos I can’t believe that the render could be this bad compared to the editor live version.
If ypur still having problem send a small clip of the svp and I can cross check if its a project setting problem or something else.
I’d also do a test and check the playback using a different audio driver. I get different results in playback using Window Audio vs ASIO.
Finally got time to check this. Now that I have uninstalled/reinstalled everything, I have a fully functional SV. I did three different exports 1) ASP as extra files, 2) ASP as extra channels and 3) No ASP.
There is very little difference between them when listening side by side other than stereo placement. It seems that whatever was causing the quality drop between the editor and export has now been resolved following the full re-installation (using Admin rights this time which might also be a factor )
Thanks very much for your help and suggestions. I hope I can be of similar help to others as my skills in this app increase.
Wait, about the support mail : do DT people read this forum? I mean, is this meaningful to post a feature request in here? Or should I send it by mail instead?
Dreamtonics has implemented features and fixed bugs that were reported on the forums before, but they don’t respond directly so it’s hard to say how often they check; they might only pay attention during beta periods and shortly after releases, and it would be very easy for them to miss topics if they hadn’t checked in some time.
During the recent 1.8.1b1 beta period I personally sent my bug reports to the support email and got responses with a generic “thank you for your feedback” type of message, so at least with emails we know they reach Dreamtonics, whereas with forum posts it’s impossible to say for certain.
I don’t think it is an editor issue. There must be something wrong with your export settings.
See, you’ve chosen the option of extracting asp. separately (i.e. aspiration output: as extra channels) but there is just no need to do this in a common and basic mixing workflow. There is a nearly 100% possibility that if you choose no separation of asp (i.e. aspiration output: none) or something like that the output file would sound normal and nice, because this time it wouldn’t output the voice as two separate parts.
Only if you need to do some de-ess or de-click will you need to separate asp., and usually to an extra file, and finally mix the processed asp. with the other part (head aligned) to get the full vocal part. I’ve used this method hundreds of times and never had any problems.
If you just listen to the part without asp. that would definitely be strange, since all the High-Frequency sounds are removed and you just can’t hear anything clearly, just the tones remain. All you need is to add asp. back.
But it just seems strange. You said that you set asp. output to none but there’s hardly any difference. That’s much confusing. Have you tried the method above? Or have you listened to the separate asp file?
Hi @leostudiooo, This is fixed now.
As I said, there is little difference between the 3 export options now when mixed in the DAW. I can hear that there is a quality drop when omitting the ASP part of the export and understand that it needs to be there for the high frequency/sibilance part of the voice. Everything you said makes sense as there was something wrong with the install that affected the exports. Now I have fully reinstalled the issue has disappeared.
Thanks for taking the time to respond