add more faqs

faq/colab-disconnect.md

@@ -1,3 +1,3 @@
-# Is Google Colab Operational?
+# Situation with Google Colab
 
 Unfortunately, Google Colab is currently unavailable as Google has taken measures to restrict access to Colab services, including Applio. We kindly request that you refrain from asking this question in the future as the situation remains unchanged.

faq/epoch.md

@@ -0,0 +1,2 @@
+# Epoch
+An Epoch is the number of iterations performed during training to complete one full cycle of your dataset. For example, if you have a dataset of 200 audio samples and you set a batch_size of 10, 10 audio samples will be processed in each iteration. To process all 200 audio samples, you will need to perform 20 iterations in total. This complete cycle is referred to as one epoch.

faq/pitch-extraction.md

@@ -0,0 +1,17 @@
+# Pitch extraction differences
+
+- **pm:** It provides a quick but inefficient result and is less faithful to the voice model.
+
+- **Harvest:** It offers a better replication of tones from our Acapella but is slower and often gives errors initially.
+
+- **Dio:** Possibly used for moments when our Acapella has rapid word pronunciation, such as in rap.
+
+- **Crepe:** It provides better word pronunciation results.
+
+- **Crepe-tiny:** Similar to crepe, it processes faster but with less aggressiveness.
+
+- **Mangio-crepe:** It provides better results in terms of breaths and performs the same as crepe, but it is slower and sometimes yields poorer results.
+
+- **Mangio-crepe-tiny:** Similar to mangio-crepe, it's faster but with less aggressiveness.
+
+- **Rmvpe:** It is a combination of pm and crepe but is fast for both training and inference. It is the most faithful to the voice model (currently the best option).

faq/tensorboard.md

@@ -0,0 +1,2 @@
+# Tensorboard
+Tensorboard is a series of graphs where we can monitor the progress of our model during training, but there are many graphs. We are only interested in the graph called 'g/total'. You can find this by clicking on 'inactive' and selecting 'scalars'. Then, go to the last page, where you will find it in the last graph.