Wiki-Quellcode von Install OpenVoice v2 Locally - Instantly Clone Voice with AI

Zuletzt geändert von René Schmidt am 2025/03/26 14:55

author	version	line-number	content
		1	conda create -n openvoice python=3.9
		2	conda activate openvoice
		3
		4
		5	git clone [[https:~~/~~/github.com/myshell-ai/openVoice.git>>https://github.com/myshell-ai/openVoice.git]]
		6
		7	cd openVoice
		8	pip install -e .
		9	mkdir checkpoints_v2
		10	cd checkpoints_v2
		11	wget [[https:~~/~~/myshell-public-repo-host.s3.amazonaws.com/openvoice/checkpoints_v2_0417.zip>>https://myshell-public-repo-host.s3.amazonaws.com/openvoice/checkpoints_v2_0417.zip]]
		12
		13	unzip checkpoints_v2_0417.zip
		14
		15
		16	mv checkpoints_v2 ~~/OpenVoice
		17
		18	cd ..
		19
		20
		21
		22
		23
		24
		25	pip install git+https:~/~/github.com/cocktailpeanut/ov2.git
		26
		27
		28
		29	cd ..
		30	pip install git+https:~/~/github.com/myshell-ai/MeloTTS.git
		31	python -m unidic download
		32	cd resources
		33	python3
		34
		35
		36
		37	import os
		38
		39	import torch
		40
		41	from openvoice import se_extractor
		42
		43	from openvoice.api import ToneColorConverter
		44
		45	ckpt_converter = 'checkpoints_v2/converter'
		46
		47	device = "cuda:0" if torch.cuda.is_available() else "cpu"
		48
		49	output_dir = 'outputs_v2'
		50
		51	tone_color_converter = ToneColorConverter(f' {ckpt_converter} /config.json', device=device)
		52
		53	tone_color_converter.load_ckpt(f' {ckpt_converter} /checkpoint.pth')
		54
		55	os.makedirs (output_dir, exist_ok=True)
		56
		57	reference_speaker = 'resources/example_reference.mp3'
		58
		59	target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, vad=False)
		60
		61	from melo.api import TTS
		62
		63	src_path = f' {output_dir} /tmp.wav'
		64
		65	speed = 1.0
		66
		67
		68
		69
		70

Anwendungen