I have been wanting to get the English voicebank for Tsurumaki Maki but I saw the issues about the memory leak, so it appears there are some different versions for the English version, there is complete, standard and AI version I believe , and I have to use the translate to see these versions and every time it opens a new page so I am just wondering if it is better to get the Standard version since it seems this version is not affected by the memory leak bug, and is the complete version both versions or is this a version that includes the voice bank made for a different program than Synth V? I have the pro studio version and want to get both standard and AI , or is there a way to just use the AI version as standard? Anyway I am a little bit confused and have not had Studio pro version for a while with the AI and I have not used it yet at all since I only have ever used the Elanor Lite version and it does not have AI, I pre ordered the Elanor voice bank in the bundle , anyway just wondering if someone could kind of clear up for me the different versions and what the state is of that memory leak, I think for now I am going to try out the lite version of Maki .
I never encountered the problem so I can’t confirm for certain, but I believe the memory leak issue was fixed within a couple days of her release.
Standard and AI are very different technologies.
Standard voices use individual samples and what is called the “concatenation method”, in which many audio files are recorded for a wide variety of sounds, at various pitches.
When the user enters notes and phonemes into the editor, the software loads each of the appropriate audio files, pitch-shifts them, and stitches them together to try and produce the desired sound.
Since Standard voices contain all the source audio that is used to construct the results they tend to be quite large on disk. Most SynthV Standard voices are a few hundred megabytes, but in other engines voices can be gigabytes in size.
AI is very different. Instead of recording each phoneme individually, a machine learning algorithm is provided with songs sung by the voice provider.
The algorithm then analyzes the way those songs are sung, and over time starts to build a profile of how the voice provider sounds and sings.
Once enough data has been analyzed, the software can use the profile to generate sounds matching the notes and pitches input by the user.
In contrast to Standard voices, AI does not include the original samples and the download tends to be very small. The result of the machine learning analysis is basically a new algorithm which can be used to generate new sounds based on the original, rather than stitching together the original samples.
Since the AI is “trained” with actual songs, AI voices are much better at producing a similar sound to the original voice provider than Standard ones.
Please check the feature comparison table and choose according to your personal needs and preferences.
The issue of memory leaks in the AI voicebanks has been fixed after version 1.2.6.
(The table uses Saki as an example, but the situation is similar for other voicebanks.)
(Translate with translator)
Ty, this is a great explanation of the differences!
Ty! That is great that the memory link was fixed that quickly, sorry I just got Pro a little bit a go so I am still catching up with this new version, so does the complete come with both the standard and the Ai version? I will go check that on the web site as well, I was experimenting around with the AI and the Regular lite version last night but I did not notice that rendering mode option, I think I will go try that a bit today, love this studio version of this synth, fantastic job on this program, this is my current favorite music program and has been for a while, i really think this technology is going to be as important to music as the electric guitar was to the acoustic guitar when it came out and predict in the future lots and lots of songs will use synth vocals and right now this is the best program out there for this by a long ways in my opinion!!!
For the download version, the AI voicebank and the standard voicebank need to be purchased separately:
(Translate with translator)
Ok, Ty for clearing that up!
I am going to get one of them in the next week or so, not sure which one yet I have been experimenting with the differences with the 2 lite versions and they do both sound different but I do not know if I am taking advantage of what the AI version can do yet compared to the non AI so I am going to get more familiar with the versions and the program as well and then decide which one to get first. I also have the Elanor Forte on pre-order so it will be great to use both of the voices in the same song,
Well after using the Pro studio a lot this week, I learned quite a bit about it, I just figured out how to use the parameters and the auto pitch tuning AI this week and I am very impressed with the updates that have been done to this program since the original synthV program, I love both of these features a lot! So I got Tsurumaki AI full version tonight and with Solaria coming up and the Male English voice in the works I am really looking forward to the future of this program! I have only had 1 English Female voice for a long time now and next year, possibly around January we will have the Solaria English voice being released, so with the AI Elanor coming out and Maki we could potentially have 3 voices to use in about 6 months with a Male English voice on the horizon, very much looking forward to using this and the improvements that have been done to this program are outstanding!!! This gives me so much more control now, well done!!