Update README.md

README.md

@@ -5,4 +5,20 @@ language:
 size_categories:
 - 1K<n<10K
 ---
-[Size-Category](https://github.com/akjava/Matcha-TTS-Japanese/wiki/Size-Category-%E3%82%B5%E3%82%A4%E3%82%BA%E3%82%AB%E3%83%86%E3%82%B4%E3%83%AA%E3%83%BC) - Large(500-5000):1230 files
+## TTS Training Detail
+[Size-Category](https://github.com/akjava/Matcha-TTS-Japanese/wiki/Size-Category-%E3%82%B5%E3%82%A4%E3%82%BA%E3%82%AB%E3%83%86%E3%82%B4%E3%83%AA%E3%83%BC) - Large(500-5000):1230 files
+
+
+
+## Audio Detail
+- 22050hz mono wav - for modern tts (original are 32khz mono mp3)
+- vadded 250ms (little bit long)
+- Volume-level:OK - no need to gain.
+- Noise-level:OK - no need to denoise
+## Creating Detail
+- Based [Common Voice](https://huggingface.co/papers/1912.06670)-[16_1 ja](https://huggingface.co/datasets/mozilla-foundation/common_voice_16_1)
+- Audios are vadded with [silero-vad](https://github.com/snakers4/silero-vad)
+- Text are cleaned simply by openjtalk.(this mean possible contain miss reading and miss converting)
+
+## Don't attempt to identify the identity of the speaker in the voices.
+I agree to download original **public domain** dataset.