UlaASMR | DMR V0.5

Create

UlaASMR | DMR V0.5

EnglishRVCE-GirlTTS / Realtime
razeristaken user image
razeristaken
1 year ago
👀

497

👍

4

🪄

14

Description

Half way done with DMR wooooo. anyway here another egrill voice Info ~20 minutes - Sample Rate: 32k - Pretrain: DMR V0.5 - Batch Size: 8 - Hop Length: 64

Comments

razeristaken user image
razeristaken
1 year ago

Half way done with DMR wooooo. anyway here another egrill voice # **Info:** - **Dataset:** ~20 minutes - **Sample Rate:** 32k - **Pretrain:** DMR V0.5 - **Batch Size:** 8 - **Hop Length:** 64 📥 **Download:** https://huggingface.co/Razer112/UlaASMR/resolve/main/Ula.zip?download=true

razeristaken user image
razeristaken
1 year ago

So i lowered the tune by 1 and it sounds more like her to me but idk

YouTubeRocks76 user image
YouTubeRocks76
1 year ago

crepe mid rmvpe peak

razeristaken user image
razeristaken
1 year ago

Crepe with a 64 hop length and rmvpe are the same

razeristaken user image
razeristaken
1 year ago

Crepe with a hop length of 32 and lower is better than rmvpe

razeristaken user image
razeristaken
1 year ago

Crepe is also just better with softer voices

litsa_the_dancer user image
litsa_the_dancer
1 year ago

just test both and see for urself

razeristaken user image
razeristaken
1 year ago

I have. So far its give me either same results as rmvpe or slightly better

litsa_the_dancer user image
litsa_the_dancer
1 year ago

give or take yea

Add a comment

Samples

New
Classic
1. Singing
Male
English
2. Singing
Female
English
3. Singing (Dry)
Female
English
4. Singing (High)
Female
English
5. Singing 2
Male
English
6. Singing (Dry)
Male
English
7. Singing (Dry, High)
Male
English

Pitch

Selected Audio
Selected Audio