Updated Home (markdown)

Home.md

@@ -1,22 +1,2 @@
 Welcome to the deep-learning-german-tts wiki!
 
-# Motivation - or why to share my private voice for mandkind
-I'm a huge fan of smart assistant systems like amazon alexa or google home on the technical side. But i can't ignore doubts on privacy. So i thought: "Someone should build an alternative for this". Beside nlu (natural language understanding) and business logic which can run locally even on smaller devices there are two aspects that currently needs cloud resources for working in acceptable quality and performance.
-
-License for 
-Please don't use for eval.
-
-# STT (speech to text) and TTS (text to speech)
-## STT
-For stt you need language specific huge datasets in many different quality levels, speaker genders, ages and dialects. This is where mozilla common voice is great ([[https://voice.mozilla.org/de]]) and deepspeech ([[https://github.com/mozilla/DeepSpeech]]) for offline voice recognition. I have read more than 13 hours phrases in german language and contribute this to public.
-
-## TTS
-When i comes to tts voices in german language there is really a limited number of options. Often the quality isn't that good, it's expensive or it's not free for use because of license restrictions.
-For tts you need a voice reading several hours of audio in a really good quality without random noise or reverb. 
-I thougth i'm using many free and opensource software components so why not contribute your own voice for free usage.
-So i've read and read and read, ... .
-And finally after months of reading i can share my ljspeech based dataset for tts generation.
-> Could be used with tacotron version 1 or 2.
-
-# Do i worry about someone missapling my voice? 
-Yes. But i'm hoping the best. So please don't use for eval :-).