firstpixel

/

F5-TTS-pt-br

like 19

There are instructions on the model card, you need to create your txt file, with the emotions and speakers, can have multiple speakers(different voices).
Also need to create your mp3 files with tones of emotion to follow, add the mp3 with speaker name and emotion as the sample bellow, add the reference audio to a ref_audios/ folder.
Add emotion to the audios, you can record changing tones and speed, at least 6 to 8 seconds on the reference audios.

input_text.txt
[speaker:speaker1, emotion:happy] Oi pessoal! Bom dia, que dia maravilhoso!
[speaker:speaker1, emotion:sad] Meu deus, só podia ser notícia ruim, não sei nem o que pensar.. estou perdido.
[speaker:speaker1, emotion:angry] Porra! Porque você fez isso? Você tá maluco? tá doido?

Adjust this to your files.
speaker_emotion_refs = {
("speaker1", "happy"): "ref_audios/speaker1_happy.wav",
("speaker1", "sad"): "ref_audios/speaker1_sad.wav",
("speaker1", "angry"): "ref_audios/speaker1_angry.wav",
}