NeMo VS stable-diffusion

I don't test WaveRNN but from the ones that I know the best that is open source is FastPitch. And it's easy to use, here is the tutorial for voice cloning.

Nowadays... bigger repos like https://github.com/NVIDIA/NeMo are all pytorch, lots of work also published by Meta and Microsoft is all torch. I check new work on GitHub all the time and I haven't seen a Tensorflow repo in years except one.

- https://github.com/NVIDIA/NeMo

Can’t you use the NeMo model and just strip the punctuation from the output again if you don’t want it? You can also fine tune the the model with capitalization only if you look at the examples https://github.com/NVIDIA/NeMo/blob/stable/tutorials/nlp/Punctuation_and_Capitalization.ipynb The capitalization and punctuation are annotated separately (U indicates that the word should be upper cased, and O - no capitalization ). The model seems to be a token level classifier not seq to seq so there should also be a way to get just the capitalization part but you would have to look into the model as it’s not shown in the examples.

I think there's been talk to do speaker diarization with whisper-asr-webservice[0] which is also written in python and should be able to make use of goodies such as pyannote-audio, py-webrtcvad, etc.

Whisper is great but at the point we get to kludging various things together it starts to make more sense to use something like Nvidia NeMo[1] which was built with all of this in mind and more

[0] - https://github.com/ahmetoner/whisper-asr-webservice

[1] - https://github.com/NVIDIA/NeMo

[커먼보이스 전자우편](mailto:[email protected]) || Common Voice || Korean Language Homepage || FAQs || Speaking Aloud and Reviewing Recordings || Sentence Collector || NVidia/NeMo

The final security grain was added with speech-to-text anti-spoofing built on QuartzNet from the Nemo framework. This model provides a decent quality user experience and is suitable for real-time scenarios. To measure how close what the person says to what the system expects, requires calculation of the Levenshtein distance between them.

Stable Diffusion: It is based on a kind of diffusion model called a latent diffusion model, which is trained to remove noise from images in an iterative process. It is one of the first text-to-image models that can run on consumer hardware and has its code and model weights publicly available.

Which is a 1-click install of Stable Diffusion with an alternative web interface. You can choose a different approach but this one is pretty simple and I am new to this stuff.

an invisible watermarking of the outputs, to help viewers identify the images as machine-generated.

It sounds like you just want to set up Stable Diffusion to run locally. I don't think your computer's specs will be able to do it. You need a graphics card with a decent amount of VRAM. Stable diffusion is in Python as is almost every AI open source project I've seen. If you can get your hands on a system with an Nvidia RTX card with as much VRAM as possible, you're in business. I have an RTX 3060 with 12 gigs of VRAM and I can run stable diffusion and a whole variety of open source LLMs as well as other projects like face swap, Roop, tortoise TTS, sadtalker, etc...

Use specific GPU on multi GPU systems · Issue #87 · CompVis/stable-diffusion · GitHub

I dealt with this in one of my posts. At least SD 1.1 till 1.5 are all trained on a batch size of 2048. The version pretty much everyone uses (1.5) is first pretrained at a resolution of 256x256 for 237K steps on laion2B-en, at the end of those training steps it will have seen roughly 500M images in laion2B-en. After that it is pre-trained for 194K steps on laion-high-resolution at a resolution of 512x512, which is a subset of 170M images from laion5B. Finally it is trained for 1.110K steps on LAION aesthetic v2 5+. This is easily verified by taking a glance at the model card of SD 1.5. Though that one doesn't specify for part of the training exactly which aesthetic set was used for part of the training, for that you have to look at the CompVis github repo. Thus at the end of it all both the most recent images and the majority of images will have come from LAION aesthetic v2 5+ (seeing every image approx 4 times). Realistically a lot of the weights obtained from pretraining on 2B will have been lost, and only provided a good starting point for the weights.

stable diffusion · CompVis/stable-diffusion@2ff270f · GitHub