diff --git a/assets/applio_bot.png b/assets/applio_bot.png index 290668da..e02b0d7b 100644 Binary files a/assets/applio_bot.png and b/assets/applio_bot.png differ diff --git a/faq/Installation/Compiled.md b/faq/Installation/Compiled.md index 1142f062..6811feaf 100644 --- a/faq/Installation/Compiled.md +++ b/faq/Installation/Compiled.md @@ -5,7 +5,7 @@ order: B --- # Precompiled -If you encounter an error during execution, you can try the [precompiled](https://huggingface.co/IAHispano/Applio/resolve/main/Applio%20V3%20Precompiled/ApplioV3.0.6.zip?download=true) version of Applio. +If you encounter an error during execution, you can try the [precompiled](https://huggingface.co/IAHispano/Applio/resolve/main/Compiled/ApplioV3.0.6.zip) version of Applio. - Download the zip, - Extract it diff --git a/faq/rvc/rvc.md b/faq/rvc/rvc.md index ef2eba1a..c3df4f9d 100644 --- a/faq/rvc/rvc.md +++ b/faq/rvc/rvc.md @@ -46,12 +46,21 @@ Hop Length is used for tonal fluctuations. Lower values are more precise but slo ==- Batch Size The batch size is the amount of GPU that will be used to train the model. The larger the batch size, the shorter the training duration. It is recommended to use multiples of 4 as the batch size, although the most common is 8, as it gives the AI time to learn correctly without rushing. ==- +==- Inference +is the process where an audio is transformed by the voice model. +==- ==- Search Feature Ratio It is the one in charge of controlling the index, the larger the ratio, the more single the dataset but it can result in artifacts, so it is better to leave it as it is by default. ==- ==- Pitch The pitch is used to adjust the tone of the model so that it can match the original voice and avoid artifacts, for men it is - and women it is +. ==- +==- Artifacting +is when the inference output sounds robotic, distorted, with background noise and with fails when trying to modulate words. +==- +==- Breath filtering +Applies respiration filtering to the results, the value represents the filter radius and respiration reduction to avoid artifacts. +==- ==- Split Audio Basically cuts the audio into parts to make the inference by parts and then joins them together. ==- @@ -61,3 +70,14 @@ It is a model trained with several sets of long-duration audios that will serve ==- Overtraining It’s when the TensorBoard graph starts to rise and never comes back down. An overtrained model will sound robotic, muffled, and won’t be able to articulate words well. For more information check the [Tensorboard section](/get-started\tensorboard.md). ==- +==- G and D +They are responsible for storing the training data of the model. +Generative learns to replicate results similar to the original. +Discriminator tries to distinguish real data from those created by the generator. +==- +==- What are the requirements for RVC/Applio? +The minimum required for local training is an Nvidia RTX series 20 graphics card with 8GB of VRAM to train models, in the case of inference you only need to have a decent CPU and at least 4GB of VRAM. If you have a Celeron processor, it would be better to look at [other alternatives](/get-started\Alternatives.md). +==- +==- Can I use RVC/Applio on a Mac? +Yes, but only for inference. besides, the installation should be done as if it were on Linux. +==- \ No newline at end of file diff --git a/get-started/Installation.md b/get-started/Installation.md index 905719d1..7cf1008d 100644 --- a/get-started/Installation.md +++ b/get-started/Installation.md @@ -8,5 +8,9 @@ icon: desktop-download Installing applio is a simple process, just download the .bat file of the latest version **[via the official repository](https://github.com/IAHispano/Applio/releases)**, extract the zip file and run the `run-install.bat (Windows) or run-install-sh (Linux)`, wait for it to download the necessary content and run the `run-applio.bat or run-applio.sh` file. !!!info Info -If you encounter an error during execution, you can try the [precompiled](https://huggingface.co/IAHispano/Applio/resolve/main/Applio%20V3%20Precompiled/ApplioV3.0.6.zip?download=true) version of Applio. +If you encounter an error during execution, you can try the [precompiled](https://huggingface.co/IAHispano/Applio/resolve/main/Compiled/ApplioV3.0.6.zip) version of Applio. +!!! + +!!!primary How to update Applio? +To update, it is necessary to delete the current Applio folder and reinstall it. Make sure to save your audios and models before deleting the current folder. !!! \ No newline at end of file diff --git a/get-started/inferencing.md b/get-started/inferencing.md index 170177b7..fd176717 100644 --- a/get-started/inferencing.md +++ b/get-started/inferencing.md @@ -26,3 +26,9 @@ Downloading files from mega may fail!, this is due to malfunction of the depende - **Recommendation:** having a **'clean acapella'** helps to get better results. but if the model downloaded in its name mentions the words **rvmpe, crepe or mangio-crepe** click on the **advanced options box** and select the right method, this will help to have more acurrate pitch control. +!!!primary What should I do if my output audio sounds robotic? + - Look for better quality audio + - Your voice model needs more [training](/get-started\training.md) or is [overtraining](/get-started\tensorboard.md). + - Remove the reverb, double vocals and noise from your acapella, check the [UVR 5 guide](/Audio\UVR.md) or [MCSEP guide](/Audio\MVSEP.md). + - The dataset of your model contained noise, you need to [clean the dataset](/Datasets\Create-Datasets.md). +!!! \ No newline at end of file diff --git a/voice-models/Bot.md b/voice-models/Bot.md index 0b21eb48..5f0579cd 100644 --- a/voice-models/Bot.md +++ b/voice-models/Bot.md @@ -12,6 +12,6 @@ The bot is capable of searching models within a database of over +500k functiona ## How to use Applio Bot -Type the command /search, enter the model name you want to search for and the type of technology you want **RVC or Kits.AI** and you’re done. +Type the command /search, enter the model name you want to search for and you’re done. ![](../assets/Applio_Bot.png)