456 lines
20 KiB
Python
456 lines
20 KiB
Python
# Usage: .venv/bin/python convert_onnx.py --model model_assets/koharune-ami/koharune-ami.safetensors
|
|
# .venv/bin/python convert_onnx.py --model model_assets/ (All models in the directory will be converted)
|
|
|
|
# https://github.com/tuna2134/sbv2-api/blob/main/scripts/convert/convert_model.py を参考に実装した
|
|
#
|
|
# MIT License
|
|
# Copyright (c) 2024 tuna2134
|
|
# Copyright (c) 2024-2025 tsukumi
|
|
#
|
|
# Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
# of this software and associated documentation files (the "Software"), to deal
|
|
# in the Software without restriction, including without limitation the rights
|
|
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
# copies of the Software, and to permit persons to whom the Software is
|
|
# furnished to do so, subject to the following conditions:
|
|
#
|
|
# The above copyright notice and this permission notice shall be included in all
|
|
# copies or substantial portions of the Software.
|
|
#
|
|
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
# SOFTWARE.
|
|
|
|
import re
|
|
import time
|
|
import uuid
|
|
from argparse import ArgumentParser
|
|
from pathlib import Path
|
|
from typing import BinaryIO, cast
|
|
|
|
import onnx
|
|
import torch
|
|
from onnxsim import model_info, simplify
|
|
from rich import print
|
|
from rich.rule import Rule
|
|
from rich.style import Style
|
|
|
|
from style_bert_vits2.constants import (
|
|
DEFAULT_ASSIST_TEXT_WEIGHT,
|
|
DEFAULT_STYLE,
|
|
DEFAULT_STYLE_WEIGHT,
|
|
Languages,
|
|
)
|
|
from style_bert_vits2.models.infer import get_text
|
|
from style_bert_vits2.models.models import SynthesizerTrn
|
|
from style_bert_vits2.models.models_jp_extra import (
|
|
SynthesizerTrn as SynthesizerTrnJPExtra,
|
|
)
|
|
from style_bert_vits2.tts_model import TTSModel
|
|
|
|
|
|
def generate_aivm_metadata(
|
|
hyper_parameters_file: BinaryIO,
|
|
style_vectors_file: BinaryIO,
|
|
model_file_name: str,
|
|
model_uuid: uuid.UUID,
|
|
):
|
|
try:
|
|
import aivmlib
|
|
from aivmlib.schemas.aivm_manifest import ModelArchitecture
|
|
except ImportError:
|
|
raise ImportError(
|
|
"aivmlib is not installed. Please install it using `pip install aivmlib`."
|
|
)
|
|
|
|
# AIVM メタデータを生成
|
|
metadata = aivmlib.generate_aivm_metadata(
|
|
# 実際に JP-Extra かどうかはハイパーパラメータの値を元に自動判定されるので、ここでは JP-Extra を指定
|
|
ModelArchitecture.StyleBertVITS2JPExtra,
|
|
hyper_parameters_file,
|
|
style_vectors_file,
|
|
)
|
|
|
|
# モデルファイル名からエポック数とステップ数を抽出
|
|
epoch_match = re.search(r"e(\d{2,})", model_file_name) # "e" の後ろに2桁以上の数字
|
|
step_match = re.search(r"s(\d{2,})", model_file_name) # "s" の後ろに2桁以上の数字
|
|
|
|
# エポック数を設定
|
|
if epoch_match:
|
|
metadata.manifest.training_epochs = int(epoch_match.group(1))
|
|
# ステップ数を設定
|
|
if step_match:
|
|
metadata.manifest.training_steps = int(step_match.group(1))
|
|
|
|
# UUID を設定
|
|
metadata.manifest.uuid = model_uuid
|
|
|
|
return metadata
|
|
|
|
|
|
if __name__ == "__main__":
|
|
start_time = time.time()
|
|
parser = ArgumentParser()
|
|
parser.add_argument(
|
|
"--model", required=True, help="Path to the model file or directory"
|
|
)
|
|
parser.add_argument(
|
|
"--force-convert",
|
|
action="store_true",
|
|
help="Already converted models will be overwritten",
|
|
)
|
|
parser.add_argument(
|
|
"--aivm",
|
|
action="store_true",
|
|
help="Generate AIVM file from Safetensors model",
|
|
)
|
|
parser.add_argument(
|
|
"--aivmx",
|
|
action="store_true",
|
|
help="Generate AIVMX file from ONNX model",
|
|
)
|
|
args = parser.parse_args()
|
|
|
|
# --model に指定されたパスがディレクトリの時、配下にある全ての .safetensors ファイルを対象に変換する
|
|
model_paths: list[Path] = []
|
|
if Path(args.model).is_dir():
|
|
for path in Path(args.model).glob("**/*.safetensors"):
|
|
# . から始まるファイルは除外
|
|
if not path.name.startswith("."):
|
|
model_paths.append(path)
|
|
else:
|
|
model_paths.append(Path(args.model))
|
|
|
|
for model_path in model_paths:
|
|
|
|
# モデルの入出力先ファイルパスを取得
|
|
onnx_temp_model_path = model_path.parent / f"{model_path.stem}_temp.onnx"
|
|
onnx_optimized_model_path = model_path.parent / f"{model_path.stem}.onnx"
|
|
config_path = model_path.parent / "config.json"
|
|
style_vec_path = model_path.parent / "style_vectors.npy"
|
|
aivm_path = model_path.parent / f"{model_path.stem}.aivm"
|
|
aivmx_path = model_path.parent / f"{model_path.stem}.aivmx"
|
|
assert model_path.exists(), "Model file does not exist"
|
|
assert config_path.exists(), "Config file does not exist"
|
|
assert style_vec_path.exists(), "Style vector file does not exist"
|
|
assert model_path.suffix != ".onnx", "Model file is already ONNX"
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
print(f"[bold cyan]Model file:[/bold cyan] {model_path}")
|
|
print(f"[bold cyan]Config file:[/bold cyan] {config_path}")
|
|
print(f"[bold cyan]Style vector file:[/bold cyan] {style_vec_path}")
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
|
|
# すでに ONNX モデルが存在する場合、--force-convert オプションが指定されていない場合はスキップ
|
|
if onnx_optimized_model_path.exists() and not args.force_convert:
|
|
print(
|
|
f"[bold yellow]ONNX model already exists: {onnx_optimized_model_path}[/bold yellow]"
|
|
)
|
|
print(
|
|
"[bold]If you want to overwrite it, use the --force-convert option.[/bold]"
|
|
)
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
|
|
# ONNX モデルが存在しない場合、ONNX モデルを生成
|
|
else:
|
|
# PyTorch モデルを読み込む
|
|
device = "cpu"
|
|
tts_model = TTSModel(
|
|
model_path=model_path,
|
|
config_path=config_path,
|
|
style_vec_path=style_vec_path,
|
|
device=device,
|
|
)
|
|
tts_model.load()
|
|
if DEFAULT_STYLE in tts_model.style2id:
|
|
style_id = tts_model.style2id[DEFAULT_STYLE]
|
|
else:
|
|
style_id = 0 # 通常デフォルトスタイルのインデックスは 0 になる
|
|
assert tts_model.net_g is not None, "Model is not loaded"
|
|
|
|
# 音声合成に必要な BERT 特徴量・音素列・アクセント列・言語 ID を取得
|
|
# JP-Extra モデルアーキテクチャの場合、bert (中国語の BERT 特徴量) や en_bert (英語の BERT 特徴量) は
|
|
# torch.zeros() で適当に埋められており、推論には ja_bert (日本語の BERT 特徴量) のみが使用される
|
|
bert, ja_bert, en_bert, phones, tones, lang_ids = get_text(
|
|
"今日はいい天気ですね。",
|
|
Languages.JP,
|
|
tts_model.hyper_parameters,
|
|
device,
|
|
assist_text=None,
|
|
assist_text_weight=DEFAULT_ASSIST_TEXT_WEIGHT,
|
|
given_phone=None,
|
|
given_tone=None,
|
|
)
|
|
|
|
# スタイルベクトルを取得
|
|
style_vector = tts_model.get_style_vector(style_id, DEFAULT_STYLE_WEIGHT)
|
|
|
|
# モデルの入力を作成
|
|
x_tst = phones.to(device).unsqueeze(0)
|
|
tones = tones.to(device).unsqueeze(0)
|
|
lang_ids = lang_ids.to(device).unsqueeze(0)
|
|
bert = bert.to(device).unsqueeze(0)
|
|
ja_bert = ja_bert.to(device).unsqueeze(0)
|
|
en_bert = en_bert.to(device).unsqueeze(0)
|
|
x_tst_lengths = torch.LongTensor([phones.size(0)]).to(device)
|
|
style_vec_tensor = torch.from_numpy(style_vector).to(device).unsqueeze(0)
|
|
sid = 0
|
|
sid_tensor = torch.LongTensor([sid]).to(device)
|
|
length_scale = torch.tensor(1.0)
|
|
sdp_ratio = torch.tensor(0.0)
|
|
noise_scale = torch.tensor(0.667)
|
|
noise_scale_w = torch.tensor(0.8)
|
|
|
|
# JP-Extra モデルアーキテクチャ向けの ONNX 変換ロジック
|
|
if isinstance(tts_model.net_g, SynthesizerTrnJPExtra):
|
|
|
|
# SynthesizerTrnJPExtra の forward メソッドをオーバーライド
|
|
def forward_jp_extra(
|
|
x: torch.Tensor,
|
|
x_lengths: torch.Tensor,
|
|
sid: torch.Tensor,
|
|
tone: torch.Tensor,
|
|
language: torch.Tensor,
|
|
bert: torch.Tensor,
|
|
style_vec: torch.Tensor,
|
|
length_scale: float = 1.0,
|
|
sdp_ratio: float = 0.0,
|
|
noise_scale: float = 0.667,
|
|
noise_scale_w: float = 0.8,
|
|
) -> tuple[
|
|
torch.Tensor, torch.Tensor, torch.Tensor, tuple[torch.Tensor, ...]
|
|
]:
|
|
return cast(SynthesizerTrnJPExtra, tts_model.net_g).infer(
|
|
x,
|
|
x_lengths,
|
|
sid,
|
|
tone,
|
|
language,
|
|
bert,
|
|
style_vec,
|
|
length_scale=length_scale,
|
|
sdp_ratio=sdp_ratio,
|
|
noise_scale=noise_scale,
|
|
noise_scale_w=noise_scale_w,
|
|
)
|
|
|
|
tts_model.net_g.forward = forward_jp_extra # type: ignore
|
|
|
|
# モデルを ONNX に変換
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
print(
|
|
"[bold cyan]Exporting ONNX model... (Architecture: JP-Extra)[/bold cyan]"
|
|
)
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
export_start_time = time.time()
|
|
torch.onnx.export(
|
|
model=tts_model.net_g,
|
|
args=(
|
|
x_tst,
|
|
x_tst_lengths,
|
|
sid_tensor,
|
|
tones,
|
|
lang_ids,
|
|
ja_bert,
|
|
style_vec_tensor,
|
|
length_scale,
|
|
sdp_ratio,
|
|
noise_scale,
|
|
noise_scale_w,
|
|
),
|
|
f=str(onnx_temp_model_path),
|
|
verbose=False,
|
|
input_names=[
|
|
"x_tst",
|
|
"x_tst_lengths",
|
|
"sid",
|
|
"tones",
|
|
"language",
|
|
"bert",
|
|
"style_vec",
|
|
"length_scale",
|
|
"sdp_ratio",
|
|
"noise_scale",
|
|
"noise_scale_w",
|
|
],
|
|
output_names=["output"],
|
|
dynamic_axes={
|
|
"x_tst": {0: "batch_size", 1: "x_tst_max_length"},
|
|
"x_tst_lengths": {0: "batch_size"},
|
|
"sid": {0: "batch_size"},
|
|
"tones": {0: "batch_size", 1: "x_tst_max_length"},
|
|
"language": {0: "batch_size", 1: "x_tst_max_length"},
|
|
"bert": {0: "batch_size", 2: "x_tst_max_length"},
|
|
"style_vec": {0: "batch_size"},
|
|
},
|
|
)
|
|
print(
|
|
f"[bold green]ONNX model exported to {onnx_temp_model_path} ({time.time() - export_start_time:.2f}s)[/bold green]"
|
|
)
|
|
|
|
# 非 JP-Extra モデルアーキテクチャ向けの ONNX 変換ロジック
|
|
else:
|
|
|
|
# SynthesizerTrn の forward メソッドをオーバーライド
|
|
def forward_non_jp_extra(
|
|
x: torch.Tensor,
|
|
x_lengths: torch.Tensor,
|
|
sid: torch.Tensor,
|
|
tone: torch.Tensor,
|
|
language: torch.Tensor,
|
|
bert: torch.Tensor,
|
|
ja_bert: torch.Tensor,
|
|
en_bert: torch.Tensor,
|
|
style_vec: torch.Tensor,
|
|
length_scale: float = 1.0,
|
|
sdp_ratio: float = 0.0,
|
|
noise_scale: float = 0.667,
|
|
noise_scale_w: float = 0.8,
|
|
) -> tuple[
|
|
torch.Tensor, torch.Tensor, torch.Tensor, tuple[torch.Tensor, ...]
|
|
]:
|
|
return cast(SynthesizerTrn, tts_model.net_g).infer(
|
|
x,
|
|
x_lengths,
|
|
sid,
|
|
tone,
|
|
language,
|
|
bert,
|
|
ja_bert,
|
|
en_bert,
|
|
style_vec,
|
|
length_scale=length_scale,
|
|
sdp_ratio=sdp_ratio,
|
|
noise_scale=noise_scale,
|
|
noise_scale_w=noise_scale_w,
|
|
)
|
|
|
|
tts_model.net_g.forward = forward_non_jp_extra # type: ignore
|
|
|
|
# モデルを ONNX に変換
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
print(
|
|
"[bold cyan]Exporting ONNX model... (Architecture: Non-JP-Extra)[/bold cyan]"
|
|
)
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
export_start_time = time.time()
|
|
torch.onnx.export(
|
|
model=tts_model.net_g,
|
|
args=(
|
|
x_tst,
|
|
x_tst_lengths,
|
|
sid_tensor,
|
|
tones,
|
|
lang_ids,
|
|
bert,
|
|
ja_bert,
|
|
en_bert,
|
|
style_vec_tensor,
|
|
length_scale,
|
|
sdp_ratio,
|
|
noise_scale,
|
|
noise_scale_w,
|
|
),
|
|
f=str(onnx_temp_model_path),
|
|
verbose=False,
|
|
input_names=[
|
|
"x_tst",
|
|
"x_tst_lengths",
|
|
"sid",
|
|
"tones",
|
|
"language",
|
|
"bert",
|
|
"ja_bert",
|
|
"en_bert",
|
|
"style_vec",
|
|
"length_scale",
|
|
"sdp_ratio",
|
|
"noise_scale",
|
|
"noise_scale_w",
|
|
],
|
|
output_names=["output"],
|
|
dynamic_axes={
|
|
"x_tst": {0: "batch_size", 1: "x_tst_max_length"},
|
|
"x_tst_lengths": {0: "batch_size"},
|
|
"sid": {0: "batch_size"},
|
|
"tones": {0: "batch_size", 1: "x_tst_max_length"},
|
|
"language": {0: "batch_size", 1: "x_tst_max_length"},
|
|
"bert": {0: "batch_size", 2: "x_tst_max_length"},
|
|
"ja_bert": {0: "batch_size", 2: "x_tst_max_length"},
|
|
"en_bert": {0: "batch_size", 2: "x_tst_max_length"},
|
|
"style_vec": {0: "batch_size"},
|
|
},
|
|
)
|
|
print(
|
|
f"[bold green]ONNX model exported to {onnx_temp_model_path} ({time.time() - export_start_time:.2f}s)[/bold green]"
|
|
)
|
|
|
|
# ONNX モデルを最適化
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
print("[bold cyan]Optimizing ONNX model...[/bold cyan]")
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
optimize_start_time = time.time()
|
|
onnx_model = onnx.load(onnx_temp_model_path)
|
|
simplified_onnx_model, check = simplify(onnx_model)
|
|
onnx.save(simplified_onnx_model, onnx_optimized_model_path)
|
|
print(
|
|
f"[bold green]ONNX model optimized and saved to {onnx_optimized_model_path} ({time.time() - optimize_start_time:.2f}s)[/bold green]"
|
|
)
|
|
print(
|
|
f"[bold]Total Time: {time.time() - start_time:.2f}s / "
|
|
f"Size: {onnx_temp_model_path.stat().st_size / 1000 / 1000:.2f}MB -> "
|
|
f"{onnx_optimized_model_path.stat().st_size / 1000 / 1000:.2f}MB[/bold]"
|
|
)
|
|
onnx_temp_model_path.unlink()
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
print("[bold cyan]Optimized model info:[/bold cyan]")
|
|
model_info.print_simplifying_info(onnx_model, simplified_onnx_model)
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
|
|
# AIVM/AIVMX ファイルを生成
|
|
if args.aivm or args.aivmx:
|
|
try:
|
|
import aivmlib
|
|
except ImportError:
|
|
raise ImportError(
|
|
"aivmlib is not installed. Please install it using `pip install aivmlib`."
|
|
)
|
|
|
|
# 共通の UUID を生成
|
|
model_uuid = uuid.uuid4()
|
|
|
|
# AIVM メタデータを生成
|
|
with config_path.open("rb") as hyper_parameters_file:
|
|
with style_vec_path.open("rb") as style_vectors_file:
|
|
aivm_metadata = generate_aivm_metadata(
|
|
hyper_parameters_file,
|
|
style_vectors_file,
|
|
model_path.name,
|
|
model_uuid,
|
|
)
|
|
|
|
# AIVM ファイルを生成
|
|
if args.aivm and (not aivm_path.exists() or args.force_convert):
|
|
print("[bold cyan]Generating AIVM file...[/bold cyan]")
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
with model_path.open("rb") as safetensors_file:
|
|
new_aivm_file_content = aivmlib.write_aivm_metadata(safetensors_file, aivm_metadata) # fmt: skip
|
|
with aivm_path.open("wb") as f:
|
|
f.write(new_aivm_file_content)
|
|
print(f"[bold green]Generated AIVM file: {aivm_path}[/bold green]")
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
|
|
# AIVMX ファイルを生成
|
|
if args.aivmx and (not aivmx_path.exists() or args.force_convert):
|
|
print("[bold cyan]Generating AIVMX file...[/bold cyan]")
|
|
print(Rule(characters="=", style=Style(color="blue")))
|
|
with onnx_optimized_model_path.open("rb") as onnx_file:
|
|
new_aivmx_file_content = aivmlib.write_aivmx_metadata(onnx_file, aivm_metadata) # fmt: skip
|
|
with aivmx_path.open("wb") as f:
|
|
f.write(new_aivmx_file_content)
|
|
print(f"[bold green]Generated AIVMX file: {aivmx_path}[/bold green]")
|
|
print(Rule(characters="=", style=Style(color="blue")))
|