Now that we have everything...... (feature suggestion: Vocal sample to SVP transcription)

Now that we have everything and coders like nothing more than a challenge, wouldn’t it be cool if we could sing into synth V and produce la-las as we can now with midi? With the “final” frontier being the ability to recognize what is being sung and lay down lyrics instead of la-las. This would reduce the workflow tremendously and I don’t care how much cpu something like that would require.

Kanru and Co, you know what to do. :wink:


i actually think theres a user made plugin for something like that but i completely forgot the name of it, i could try asking around if your curious?


You might be thinking of RealVoice which imports pitch and timing data from vocal recordings to SynthV Studio, but still involves many manual steps (GitHub - hataori-p/real-voice: Scripts for working with a real voice in Synthesizer V Studio Pro)

I believe the suggestion here is direct lyric and pitch transcription into SVP format, which would be a neat feature but I’m not sure it would be worth diverting efforts from the core functionality of the product (unless Dreamtonics is doing well enough to hire on new staff, I suppose).


Ooooh Thanks, I’ll try that. Man I need to get out more.

I do already have Praat. So I’ll give it a go

I got lost with Praat and gave up.

I too think this is a good idea and a worthy effort for the developers. Ease of input and creation will hopefully be a priority for the team.

I am new to Synth V and was showing it off to a fellow music producer and he made the same suggestion as the original poster. The ability to be able to sing in a rough vocal and have the program recognize the pitch, timing and lyrics as a starting point. While I agree with Claire that this would require a huge amount of effort, I can’t help but think that this would be a huge workflow enhancement. Overall, I think improving the quality and expressiveness (adding grit, rasp and shred) of the voicebanks should be the developers primary focus, but I also think that ideas like this are great to put on their radar. With machine learning growing at its current pace, it is only a matter of time before someone develops this type of functionality.