Hello all,
I asked about an idea on another topic, Mimic II pre-trained model, about using my grandmothers voice with Mimic II
My issue is there is no way to get her to sit and record a set of phrases to do this. My solution is to use a high quality lavaliere microphone and just record conversations with her. I know that this would be a very tedious task, and involve clipping the audio into small clips and transcribe each, but I am willing to do this.
I would love to hear feedback about this, and give me ideas on mics to use, and the best software for recording also. The amount of audio that I need shouldn’t be a problem, as she lives with my aunt and can record as much as needed. I think with a high quality mic, this should be possible.
How about a processor to use this voice? Do I have to run my own Mimic II server? Or can I use the online TTS and just use a local voice?
You’ll more than likely want a gpu to run a local copy of mimic2. I’m not aware of any TTS service that lets you upload a model to use. I suspect none will anytime soon,either.
Mics all depends on budget, and the cheap ones aren’t great. There’s good budget mics, there’s great mics that aren’t low budget. Find one that works for your scenario. try renting from a local music shop maybe? Skip software since that would tie you into a computer by her non-stop, and just get a small recording device (zoom/tascam/sony/etc). Make sure it’s recording at highest quality. Digital recorders can do 24bit/48khz or more, even though you’ll end up converting to something less. You can’t recover data that’s not recorded, so be sure you get it right the first time.
Practice on yourself a few times before you try it with grandma, make it as smooth and easy for her as possible.
Thank you @brrn. Yes, I was hoping to use the mimic recording studio but use audio clips instead of a direct microphone. I have not examined it closely yet, and am not sure that is possible.