👨‍🏫 Tutorial Microsoft just dropped a open-source voice AI that actually slaps ElevenLabs and its 100% FREE

[alist1986]

wow, thank you sobra, ang galing ng demo video, human voice na ata yun,

Di pa perfect ang grammar nyan, pero sa tunog magandang pakinggan. Gamit ko yan pag-create ng storytelling audiobooks experiemnts. Ayos din pag inayos ko yung sctipt at sceenplay, scene by scene. May limits kasi yan.

 

[Emorowena]

Di pa perfect ang grammar nyan, pero sa tunog magandang pakinggan. Gamit ko yan pag-create ng storytelling audiobooks experiemnts. Ayos din pag inayos ko yung sctipt at sceenplay, scene by scene. May limits kasi yan.

sobrang limit naka 2x a min, lang limit na or mali nagawa ko.

 

[PapaEpz]

Salamat po dito lodi!!! Keep on Sharing po !!!

 

[alist1986]

sobrang limit naka 2x a min, lang limit na or mali nagawa ko.

Di mo pala nakuha yung hint ko, he he. Yung binigay ko sa'yong TTS maraming free online nyan - hindi lang dyan. Kolektahin mo yung apis nila. Di ba matagal sa Colab? Ang alam ko 12 GPU hrs/day sa inference w/out training - sulit na yon. Di kasi ako palagamit dyan. Di na ako magbabanggit ng link para iwas sisi kung mamali ako.
Sa ibang mga Chinese AIs alam ko di mahigpit pa. Sa magandang open source, ito lang yung safe gamitin na alam ko sa ngayon: OpenF5-TTS (Apache 2.0), Kokoro (Apache 2.0), or Chatterbox (MIT). To clarify the licencing sa F5-TTS, ito yung difference:

• Original F5-TTS: Created by You do not have permission to view the full content of this post. Log in or register now., this model was trained on the massive Emilia dataset. Because the training data carries restrictions, the resulting official model weights are limited to non-commercial use (CC-BY-NC-4.0).
• OpenF5-TTS: This is an independent, community-driven community checkpoint (such as the You do not have permission to view the full content of this post. Log in or register now. variant). It was built using the F5-TTS code repository but trained exclusively on permissively licensed datasets. Because of this, it is released under the Apache 2.0 license, which explicitly allows for full commercial use.

Medyo technical, pero klaro naman intindihin.
Ito yung for cpu with GUI:
You do not have permission to view the full content of this post. Log in or register now.
or this:
You do not have permission to view the full content of this post. Log in or register now.
Yung una, na-test ko pero mabagal sa'kin. Parang orpheus din siya. Pero yung ONNX version ng orpheus, pang-GPU siguro at 3B parameter na yon or try if there are GGUF-quantized versions na maliliit na models.

Kung sa licensing ka titirahin, talagang forced ka to pay or find an open source that is ok to use commercially. Gamitan mo ng You do not have permission to view the full content of this post. Log in or register now. para yung audio creation mo ay hindi manakaw ng iba, at sila naman yung tirahin mo, he he.