117 lines
3.1 KiB
Python
117 lines
3.1 KiB
Python
import click
|
|
from loguru import logger
|
|
|
|
|
|
@click.group()
|
|
def cli():
|
|
pass
|
|
|
|
|
|
@cli.command()
|
|
@click.option(
|
|
"--backend", default="crispasr", help="The backend to use for speech recognition"
|
|
)
|
|
@click.option(
|
|
"--model", default="./qwen3tts", help="The model to use for speech recognition"
|
|
)
|
|
@click.option(
|
|
"--output", default="./output.wav", help="The output file for the generated speech"
|
|
)
|
|
@click.option(
|
|
"--output-dir",
|
|
default="./outputs",
|
|
help="The output directory for the generated speech",
|
|
)
|
|
@click.option(
|
|
"--gpu-backend",
|
|
default="auto",
|
|
help="The GPU backend to use for speech recognition",
|
|
)
|
|
@click.option(
|
|
"--instruct",
|
|
default="用符合文本应读语气说",
|
|
help="The instruction to use for speech recognition",
|
|
)
|
|
@click.option(
|
|
"executable",
|
|
"--executable",
|
|
default="./crispasr",
|
|
help="The executable to use for speech recognition",
|
|
)
|
|
@click.option(
|
|
"--codec-model",
|
|
help="The codec model to use for speech recognition",
|
|
)
|
|
@click.option(
|
|
"--strict",
|
|
is_flag=True,
|
|
help="Use strict parsing mode,which just keep Chinese, English and numbers",
|
|
)
|
|
@click.argument("input_file")
|
|
def say(
|
|
backend: str,
|
|
model: str,
|
|
output: str,
|
|
input_file: str,
|
|
gpu_backend: str,
|
|
instruct: str,
|
|
executable: str,
|
|
output_dir: str,
|
|
codec_model: str,
|
|
strict: bool,
|
|
):
|
|
with open(input_file, "r", encoding="utf-8") as f:
|
|
html = f.read()
|
|
from parser import TTSHTMLParser
|
|
|
|
logger.info("解析中...")
|
|
parser = TTSHTMLParser(html)
|
|
result = parser.parse(strict=strict)
|
|
logger.debug(f"解析结果: {result}")
|
|
match backend:
|
|
case "crispasr":
|
|
from backend import CrispASRCLIBackend
|
|
|
|
tts_backend = CrispASRCLIBackend(
|
|
model=model,
|
|
executable=executable,
|
|
gpu_backend=gpu_backend,
|
|
codec_model=codec_model,
|
|
)
|
|
|
|
for voice, text in result:
|
|
logger.info(f"正在合成: voice={voice}, text={text}")
|
|
tts_backend.smart_say(
|
|
voice=voice,
|
|
text=text,
|
|
instruct=instruct,
|
|
output_file_prefix=f"{output_dir}/{voice}_{text[0:9]}",
|
|
)
|
|
case _:
|
|
raise ValueError(f"Unsupported backend: {backend}")
|
|
|
|
|
|
@cli.command()
|
|
@click.argument("input_dir")
|
|
@click.argument("output_file")
|
|
@click.option("--format", default="wav", help="The output format, either wav or mp3")
|
|
def combine(input_dir: str, output_file: str, format: str):
|
|
from files import FileHandler
|
|
|
|
handler = FileHandler(input_dir)
|
|
if format == "wav":
|
|
handler.combine_waves(output_file)
|
|
elif format == "mp3":
|
|
handler.combine_waves(output_file)
|
|
handler.wavs_to_mp3(output_file, output_file.replace(".wav", ".mp3"))
|
|
|
|
|
|
@cli.command()
|
|
@click.argument("input_file")
|
|
@click.argument("output_file")
|
|
def transform(input_file: str, output_file: str):
|
|
from files import FileHandler
|
|
|
|
handler = FileHandler(input_file)
|
|
handler.wavs_to_mp3(input_file, output_file)
|