From 2b5fb886b1c8fbacf3608caf6fd10f1ed76a799d Mon Sep 17 00:00:00 2001 From: umoubuton <127150330+umoufuton@users.noreply.github.com> Date: Wed, 5 Jul 2023 21:53:17 +0800 Subject: [PATCH 1/4] Update TTS in webUI --- edgetts/tts.py | 47 +++++++ edgetts/tts_voices.py | 306 ++++++++++++++++++++++++++++++++++++++++++ webUI.py | 185 ++++++++++++------------- 3 files changed, 448 insertions(+), 90 deletions(-) create mode 100644 edgetts/tts.py create mode 100644 edgetts/tts_voices.py diff --git a/edgetts/tts.py b/edgetts/tts.py new file mode 100644 index 0000000..03fd5bf --- /dev/null +++ b/edgetts/tts.py @@ -0,0 +1,47 @@ +import asyncio +import random +import edge_tts +from edge_tts import VoicesManager +import sys +from langdetect import detect +from langdetect import DetectorFactory + +DetectorFactory.seed = 0 + +TEXT = sys.argv[1] +LANG = detect(TEXT) if sys.argv[2] == "Auto" else sys.argv[2] +if LANG == "zh-cn" or LANG == "zh-tw": + LOCALE = LANG[:-2] + LANG[-2:].upper() +RATE = sys.argv[3] +VOLUME = sys.argv[4] +GENDER = sys.argv[5] if len(sys.argv) == 6 else None +OUTPUT_FILE = "tts.wav" + +print("Running TTS...") +print(f"Text: {TEXT}, Language: {LANG}, Gender: {GENDER}, Rate: {RATE}, Volume: {VOLUME}") + +async def _main() -> None: + voices = await VoicesManager.create() + if not GENDER is None: + if LANG == "zh-cn" or LANG == "zh-tw": + voice = voices.find(Gender=GENDER, Locale=LOCALE) + else: + voice = voices.find(Gender=GENDER, Language=LANG) + VOICE = random.choice(voice)["Name"] + print(f"Using random {LANG} voice: {VOICE}") + else: + VOICE = LANG + + communicate = edge_tts.Communicate(text = TEXT, voice = VOICE, rate = RATE, volume = VOLUME) + await communicate.save(OUTPUT_FILE) + +if __name__ == "__main__": + if sys.platform.startswith("win"): + asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy()) + asyncio.run(_main()) + else: + loop = asyncio.get_event_loop_policy().get_event_loop() + try: + loop.run_until_complete(_main()) + finally: + loop.close() \ No newline at end of file diff --git a/edgetts/tts_voices.py b/edgetts/tts_voices.py new file mode 100644 index 0000000..b59620a --- /dev/null +++ b/edgetts/tts_voices.py @@ -0,0 +1,306 @@ +#List of Supported Voices for edge_TTS +SUPPORTED_VOICES = { + 'zh-CN-XiaoxiaoNeural': 'zh-CN', + 'zh-CN-XiaoyiNeural': 'zh-CN', + 'zh-CN-YunjianNeural': 'zh-CN', + 'zh-CN-YunxiNeural': 'zh-CN', + 'zh-CN-YunxiaNeural': 'zh-CN', + 'zh-CN-YunyangNeural': 'zh-CN', + 'zh-HK-HiuGaaiNeural': 'zh-HK', + 'zh-HK-HiuMaanNeural': 'zh-HK', + 'zh-HK-WanLungNeural': 'zh-HK', + 'zh-TW-HsiaoChenNeural': 'zh-TW', + 'zh-TW-YunJheNeural': 'zh-TW', + 'zh-TW-HsiaoYuNeural': 'zh-TW', + 'af-ZA-AdriNeural': 'af-ZA', + 'af-ZA-WillemNeural': 'af-ZA', + 'am-ET-AmehaNeural': 'am-ET', + 'am-ET-MekdesNeural': 'am-ET', + 'ar-AE-FatimaNeural': 'ar-AE', + 'ar-AE-HamdanNeural': 'ar-AE', + 'ar-BH-AliNeural': 'ar-BH', + 'ar-BH-LailaNeural': 'ar-BH', + 'ar-DZ-AminaNeural': 'ar-DZ', + 'ar-DZ-IsmaelNeural': 'ar-DZ', + 'ar-EG-SalmaNeural': 'ar-EG', + 'ar-EG-ShakirNeural': 'ar-EG', + 'ar-IQ-BasselNeural': 'ar-IQ', + 'ar-IQ-RanaNeural': 'ar-IQ', + 'ar-JO-SanaNeural': 'ar-JO', + 'ar-JO-TaimNeural': 'ar-JO', + 'ar-KW-FahedNeural': 'ar-KW', + 'ar-KW-NouraNeural': 'ar-KW', + 'ar-LB-LaylaNeural': 'ar-LB', + 'ar-LB-RamiNeural': 'ar-LB', + 'ar-LY-ImanNeural': 'ar-LY', + 'ar-LY-OmarNeural': 'ar-LY', + 'ar-MA-JamalNeural': 'ar-MA', + 'ar-MA-MounaNeural': 'ar-MA', + 'ar-OM-AbdullahNeural': 'ar-OM', + 'ar-OM-AyshaNeural': 'ar-OM', + 'ar-QA-AmalNeural': 'ar-QA', + 'ar-QA-MoazNeural': 'ar-QA', + 'ar-SA-HamedNeural': 'ar-SA', + 'ar-SA-ZariyahNeural': 'ar-SA', + 'ar-SY-AmanyNeural': 'ar-SY', + 'ar-SY-LaithNeural': 'ar-SY', + 'ar-TN-HediNeural': 'ar-TN', + 'ar-TN-ReemNeural': 'ar-TN', + 'ar-YE-MaryamNeural': 'ar-YE', + 'ar-YE-SalehNeural': 'ar-YE', + 'az-AZ-BabekNeural': 'az-AZ', + 'az-AZ-BanuNeural': 'az-AZ', + 'bg-BG-BorislavNeural': 'bg-BG', + 'bg-BG-KalinaNeural': 'bg-BG', + 'bn-BD-NabanitaNeural': 'bn-BD', + 'bn-BD-PradeepNeural': 'bn-BD', + 'bn-IN-BashkarNeural': 'bn-IN', + 'bn-IN-TanishaaNeural': 'bn-IN', + 'bs-BA-GoranNeural': 'bs-BA', + 'bs-BA-VesnaNeural': 'bs-BA', + 'ca-ES-EnricNeural': 'ca-ES', + 'ca-ES-JoanaNeural': 'ca-ES', + 'cs-CZ-AntoninNeural': 'cs-CZ', + 'cs-CZ-VlastaNeural': 'cs-CZ', + 'cy-GB-AledNeural': 'cy-GB', + 'cy-GB-NiaNeural': 'cy-GB', + 'da-DK-ChristelNeural': 'da-DK', + 'da-DK-JeppeNeural': 'da-DK', + 'de-AT-IngridNeural': 'de-AT', + 'de-AT-JonasNeural': 'de-AT', + 'de-CH-JanNeural': 'de-CH', + 'de-CH-LeniNeural': 'de-CH', + 'de-DE-AmalaNeural': 'de-DE', + 'de-DE-ConradNeural': 'de-DE', + 'de-DE-KatjaNeural': 'de-DE', + 'de-DE-KillianNeural': 'de-DE', + 'el-GR-AthinaNeural': 'el-GR', + 'el-GR-NestorasNeural': 'el-GR', + 'en-AU-NatashaNeural': 'en-AU', + 'en-AU-WilliamNeural': 'en-AU', + 'en-CA-ClaraNeural': 'en-CA', + 'en-CA-LiamNeural': 'en-CA', + 'en-GB-LibbyNeural': 'en-GB', + 'en-GB-MaisieNeural': 'en-GB', + 'en-GB-RyanNeural': 'en-GB', + 'en-GB-SoniaNeural': 'en-GB', + 'en-GB-ThomasNeural': 'en-GB', + 'en-HK-SamNeural': 'en-HK', + 'en-HK-YanNeural': 'en-HK', + 'en-IE-ConnorNeural': 'en-IE', + 'en-IE-EmilyNeural': 'en-IE', + 'en-IN-NeerjaNeural': 'en-IN', + 'en-IN-PrabhatNeural': 'en-IN', + 'en-KE-AsiliaNeural': 'en-KE', + 'en-KE-ChilembaNeural': 'en-KE', + 'en-NG-AbeoNeural': 'en-NG', + 'en-NG-EzinneNeural': 'en-NG', + 'en-NZ-MitchellNeural': 'en-NZ', + 'en-NZ-MollyNeural': 'en-NZ', + 'en-PH-JamesNeural': 'en-PH', + 'en-PH-RosaNeural': 'en-PH', + 'en-SG-LunaNeural': 'en-SG', + 'en-SG-WayneNeural': 'en-SG', + 'en-TZ-ElimuNeural': 'en-TZ', + 'en-TZ-ImaniNeural': 'en-TZ', + 'en-US-AnaNeural': 'en-US', + 'en-US-AriaNeural': 'en-US', + 'en-US-ChristopherNeural': 'en-US', + 'en-US-EricNeural': 'en-US', + 'en-US-GuyNeural': 'en-US', + 'en-US-JennyNeural': 'en-US', + 'en-US-MichelleNeural': 'en-US', + 'en-ZA-LeahNeural': 'en-ZA', + 'en-ZA-LukeNeural': 'en-ZA', + 'es-AR-ElenaNeural': 'es-AR', + 'es-AR-TomasNeural': 'es-AR', + 'es-BO-MarceloNeural': 'es-BO', + 'es-BO-SofiaNeural': 'es-BO', + 'es-CL-CatalinaNeural': 'es-CL', + 'es-CL-LorenzoNeural': 'es-CL', + 'es-CO-GonzaloNeural': 'es-CO', + 'es-CO-SalomeNeural': 'es-CO', + 'es-CR-JuanNeural': 'es-CR', + 'es-CR-MariaNeural': 'es-CR', + 'es-CU-BelkysNeural': 'es-CU', + 'es-CU-ManuelNeural': 'es-CU', + 'es-DO-EmilioNeural': 'es-DO', + 'es-DO-RamonaNeural': 'es-DO', + 'es-EC-AndreaNeural': 'es-EC', + 'es-EC-LuisNeural': 'es-EC', + 'es-ES-AlvaroNeural': 'es-ES', + 'es-ES-ElviraNeural': 'es-ES', + 'es-ES-ManuelEsCUNeural': 'es-ES', + 'es-GQ-JavierNeural': 'es-GQ', + 'es-GQ-TeresaNeural': 'es-GQ', + 'es-GT-AndresNeural': 'es-GT', + 'es-GT-MartaNeural': 'es-GT', + 'es-HN-CarlosNeural': 'es-HN', + 'es-HN-KarlaNeural': 'es-HN', + 'es-MX-DaliaNeural': 'es-MX', + 'es-MX-JorgeNeural': 'es-MX', + 'es-MX-LorenzoEsCLNeural': 'es-MX', + 'es-NI-FedericoNeural': 'es-NI', + 'es-NI-YolandaNeural': 'es-NI', + 'es-PA-MargaritaNeural': 'es-PA', + 'es-PA-RobertoNeural': 'es-PA', + 'es-PE-AlexNeural': 'es-PE', + 'es-PE-CamilaNeural': 'es-PE', + 'es-PR-KarinaNeural': 'es-PR', + 'es-PR-VictorNeural': 'es-PR', + 'es-PY-MarioNeural': 'es-PY', + 'es-PY-TaniaNeural': 'es-PY', + 'es-SV-LorenaNeural': 'es-SV', + 'es-SV-RodrigoNeural': 'es-SV', + 'es-US-AlonsoNeural': 'es-US', + 'es-US-PalomaNeural': 'es-US', + 'es-UY-MateoNeural': 'es-UY', + 'es-UY-ValentinaNeural': 'es-UY', + 'es-VE-PaolaNeural': 'es-VE', + 'es-VE-SebastianNeural': 'es-VE', + 'et-EE-AnuNeural': 'et-EE', + 'et-EE-KertNeural': 'et-EE', + 'fa-IR-DilaraNeural': 'fa-IR', + 'fa-IR-FaridNeural': 'fa-IR', + 'fi-FI-HarriNeural': 'fi-FI', + 'fi-FI-NooraNeural': 'fi-FI', + 'fil-PH-AngeloNeural': 'fil-PH', + 'fil-PH-BlessicaNeural': 'fil-PH', + 'fr-BE-CharlineNeural': 'fr-BE', + 'fr-BE-GerardNeural': 'fr-BE', + 'fr-CA-AntoineNeural': 'fr-CA', + 'fr-CA-JeanNeural': 'fr-CA', + 'fr-CA-SylvieNeural': 'fr-CA', + 'fr-CH-ArianeNeural': 'fr-CH', + 'fr-CH-FabriceNeural': 'fr-CH', + 'fr-FR-DeniseNeural': 'fr-FR', + 'fr-FR-EloiseNeural': 'fr-FR', + 'fr-FR-HenriNeural': 'fr-FR', + 'ga-IE-ColmNeural': 'ga-IE', + 'ga-IE-OrlaNeural': 'ga-IE', + 'gl-ES-RoiNeural': 'gl-ES', + 'gl-ES-SabelaNeural': 'gl-ES', + 'gu-IN-DhwaniNeural': 'gu-IN', + 'gu-IN-NiranjanNeural': 'gu-IN', + 'he-IL-AvriNeural': 'he-IL', + 'he-IL-HilaNeural': 'he-IL', + 'hi-IN-MadhurNeural': 'hi-IN', + 'hi-IN-SwaraNeural': 'hi-IN', + 'hr-HR-GabrijelaNeural': 'hr-HR', + 'hr-HR-SreckoNeural': 'hr-HR', + 'hu-HU-NoemiNeural': 'hu-HU', + 'hu-HU-TamasNeural': 'hu-HU', + 'id-ID-ArdiNeural': 'id-ID', + 'id-ID-GadisNeural': 'id-ID', + 'is-IS-GudrunNeural': 'is-IS', + 'is-IS-GunnarNeural': 'is-IS', + 'it-IT-DiegoNeural': 'it-IT', + 'it-IT-ElsaNeural': 'it-IT', + 'it-IT-IsabellaNeural': 'it-IT', + 'ja-JP-KeitaNeural': 'ja-JP', + 'ja-JP-NanamiNeural': 'ja-JP', + 'jv-ID-DimasNeural': 'jv-ID', + 'jv-ID-SitiNeural': 'jv-ID', + 'ka-GE-EkaNeural': 'ka-GE', + 'ka-GE-GiorgiNeural': 'ka-GE', + 'kk-KZ-AigulNeural': 'kk-KZ', + 'kk-KZ-DauletNeural': 'kk-KZ', + 'km-KH-PisethNeural': 'km-KH', + 'km-KH-SreymomNeural': 'km-KH', + 'kn-IN-GaganNeural': 'kn-IN', + 'kn-IN-SapnaNeural': 'kn-IN', + 'ko-KR-InJoonNeural': 'ko-KR', + 'ko-KR-SunHiNeural': 'ko-KR', + 'lo-LA-ChanthavongNeural': 'lo-LA', + 'lo-LA-KeomanyNeural': 'lo-LA', + 'lt-LT-LeonasNeural': 'lt-LT', + 'lt-LT-OnaNeural': 'lt-LT', + 'lv-LV-EveritaNeural': 'lv-LV', + 'lv-LV-NilsNeural': 'lv-LV', + 'mk-MK-AleksandarNeural': 'mk-MK', + 'mk-MK-MarijaNeural': 'mk-MK', + 'ml-IN-MidhunNeural': 'ml-IN', + 'ml-IN-SobhanaNeural': 'ml-IN', + 'mn-MN-BataaNeural': 'mn-MN', + 'mn-MN-YesuiNeural': 'mn-MN', + 'mr-IN-AarohiNeural': 'mr-IN', + 'mr-IN-ManoharNeural': 'mr-IN', + 'ms-MY-OsmanNeural': 'ms-MY', + 'ms-MY-YasminNeural': 'ms-MY', + 'mt-MT-GraceNeural': 'mt-MT', + 'mt-MT-JosephNeural': 'mt-MT', + 'my-MM-NilarNeural': 'my-MM', + 'my-MM-ThihaNeural': 'my-MM', + 'nb-NO-FinnNeural': 'nb-NO', + 'nb-NO-PernilleNeural': 'nb-NO', + 'ne-NP-HemkalaNeural': 'ne-NP', + 'ne-NP-SagarNeural': 'ne-NP', + 'nl-BE-ArnaudNeural': 'nl-BE', + 'nl-BE-DenaNeural': 'nl-BE', + 'nl-NL-ColetteNeural': 'nl-NL', + 'nl-NL-FennaNeural': 'nl-NL', + 'nl-NL-MaartenNeural': 'nl-NL', + 'pl-PL-MarekNeural': 'pl-PL', + 'pl-PL-ZofiaNeural': 'pl-PL', + 'ps-AF-GulNawazNeural': 'ps-AF', + 'ps-AF-LatifaNeural': 'ps-AF', + 'pt-BR-AntonioNeural': 'pt-BR', + 'pt-BR-FranciscaNeural': 'pt-BR', + 'pt-PT-DuarteNeural': 'pt-PT', + 'pt-PT-RaquelNeural': 'pt-PT', + 'ro-RO-AlinaNeural': 'ro-RO', + 'ro-RO-EmilNeural': 'ro-RO', + 'ru-RU-DmitryNeural': 'ru-RU', + 'ru-RU-SvetlanaNeural': 'ru-RU', + 'si-LK-SameeraNeural': 'si-LK', + 'si-LK-ThiliniNeural': 'si-LK', + 'sk-SK-LukasNeural': 'sk-SK', + 'sk-SK-ViktoriaNeural': 'sk-SK', + 'sl-SI-PetraNeural': 'sl-SI', + 'sl-SI-RokNeural': 'sl-SI', + 'so-SO-MuuseNeural': 'so-SO', + 'so-SO-UbaxNeural': 'so-SO', + 'sq-AL-AnilaNeural': 'sq-AL', + 'sq-AL-IlirNeural': 'sq-AL', + 'sr-RS-NicholasNeural': 'sr-RS', + 'sr-RS-SophieNeural': 'sr-RS', + 'su-ID-JajangNeural': 'su-ID', + 'su-ID-TutiNeural': 'su-ID', + 'sv-SE-MattiasNeural': 'sv-SE', + 'sv-SE-SofieNeural': 'sv-SE', + 'sw-KE-RafikiNeural': 'sw-KE', + 'sw-KE-ZuriNeural': 'sw-KE', + 'sw-TZ-DaudiNeural': 'sw-TZ', + 'sw-TZ-RehemaNeural': 'sw-TZ', + 'ta-IN-PallaviNeural': 'ta-IN', + 'ta-IN-ValluvarNeural': 'ta-IN', + 'ta-LK-KumarNeural': 'ta-LK', + 'ta-LK-SaranyaNeural': 'ta-LK', + 'ta-MY-KaniNeural': 'ta-MY', + 'ta-MY-SuryaNeural': 'ta-MY', + 'ta-SG-AnbuNeural': 'ta-SG', + 'ta-SG-VenbaNeural': 'ta-SG', + 'te-IN-MohanNeural': 'te-IN', + 'te-IN-ShrutiNeural': 'te-IN', + 'th-TH-NiwatNeural': 'th-TH', + 'th-TH-PremwadeeNeural': 'th-TH', + 'tr-TR-AhmetNeural': 'tr-TR', + 'tr-TR-EmelNeural': 'tr-TR', + 'uk-UA-OstapNeural': 'uk-UA', + 'uk-UA-PolinaNeural': 'uk-UA', + 'ur-IN-GulNeural': 'ur-IN', + 'ur-IN-SalmanNeural': 'ur-IN', + 'ur-PK-AsadNeural': 'ur-PK', + 'ur-PK-UzmaNeural': 'ur-PK', + 'uz-UZ-MadinaNeural': 'uz-UZ', + 'uz-UZ-SardorNeural': 'uz-UZ', + 'vi-VN-HoaiMyNeural': 'vi-VN', + 'vi-VN-NamMinhNeural': 'vi-VN', + 'zu-ZA-ThandoNeural': 'zu-ZA', + 'zu-ZA-ThembaNeural': 'zu-ZA', +} + +SUPPORTED_LANGUAGES = [ + "Auto", + *SUPPORTED_VOICES.keys() +] \ No newline at end of file diff --git a/webUI.py b/webUI.py index e3a8cad..c4ded2d 100644 --- a/webUI.py +++ b/webUI.py @@ -5,7 +5,9 @@ import re import subprocess import time import traceback + from itertools import chain +from pathlib import Path # os.system("wget -P cvec/ https://huggingface.co/spaces/innnky/nanami/resolve/main/checkpoint_best_legacy_500.pt") import gradio as gr @@ -19,6 +21,7 @@ from scipy.io import wavfile from compress_model import removeOptimizer from inference.infer_tool import Svc from utils import mix_model +from edgetts.tts_voices import SUPPORTED_LANGUAGES logging.getLogger('numba').setLevel(logging.WARNING) logging.getLogger('markdown_it').setLevel(logging.WARNING) @@ -81,7 +84,6 @@ def modelAnalysis(model_path,config_path,cluster_model_path,device,enhance,diff_ device = cuda[device] if "CUDA" in device else device cluster_filepath = os.path.split(cluster_model_path.name) if cluster_model_path is not None else "no_cluster" fr = ".pkl" in cluster_filepath[1] - #model = Svc(model_path.name, config_path.name, device=device if device!="Auto" else None, cluster_model_path = cluster_model_path.name if cluster_model_path != None else "",nsf_hifigan_enhance=enhance) model = Svc(model_path.name, config_path.name, device=device if device != "Auto" else None, @@ -126,108 +128,107 @@ def modelUnload(): model = None torch.cuda.empty_cache() return sid.update(choices = [],value=""),"模型卸载完毕!" + +def vc_infer(output_format, sid, audio_path, truncated_basename, vc_transform, auto_f0, cluster_ratio, slice_db, noise_scale, pad_seconds, cl_num, lg_num, lgr_num, f0_predictor, enhancer_adaptive_key, cr_threshold, k_step, use_spk_mix, second_encoding, loudness_envelope_adjustment): + global model + _audio = model.slice_inference( + audio_path, + sid, + vc_transform, + slice_db, + cluster_ratio, + auto_f0, + noise_scale, + pad_seconds, + cl_num, + lg_num, + lgr_num, + f0_predictor, + enhancer_adaptive_key, + cr_threshold, + k_step, + use_spk_mix, + second_encoding, + loudness_envelope_adjustment + ) + model.clear_empty() + #构建保存文件的路径，并保存到results文件夹内 + str(int(time.time())) + if not os.path.exists("results"): + os.makedirs("results") + key = "auto" if auto_f0 else f"{int(vc_transform)}key" + cluster = "_" if cluster_ratio == 0 else f"_{cluster_ratio}_" + isdiffusion = "sovits" + if model.shallow_diffusion: + isdiffusion = "sovdiff" -def vc_fn(sid, input_audio, vc_transform, auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,f0_predictor,enhancer_adaptive_key,cr_threshold,k_step,use_spk_mix,second_encoding,loudness_envelope_adjustment): + if model.only_diffusion: + isdiffusion = "diff" + + output_file_name = 'result_'+truncated_basename+f'_{sid}_{key}{cluster}{isdiffusion}.{output_format}' + output_file = os.path.join("results", output_file_name) + soundfile.write(output_file, _audio, model.target_sample, format=output_format) + return output_file + +def vc_fn(sid, input_audio, output_format, vc_transform, auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,f0_predictor,enhancer_adaptive_key,cr_threshold,k_step,use_spk_mix,second_encoding,loudness_envelope_adjustment): global model try: if input_audio is None: return "You need to upload an audio", None if model is None: return "You need to upload an model", None - print(input_audio) - sampling_rate, audio = input_audio - print(audio.shape,sampling_rate) - audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32) - print(audio.dtype) + if getattr(model, 'cluster_model', None) is None and model.feature_retrieval is False: + if cluster_ratio != 0: + return "You need to upload an cluster model or feature retrieval model before assigning cluster ratio!", None + #print(input_audio) + audio, sampling_rate = soundfile.read(input_audio) + #print(audio.shape,sampling_rate) + if np.issubdtype(audio.dtype, np.integer): + audio = (audio / np.iinfo(audio.dtype).max).astype(np.float32) + #print(audio.dtype) if len(audio.shape) > 1: audio = librosa.to_mono(audio.transpose(1, 0)) - temp_path = "temp.wav" - soundfile.write(temp_path, audio, sampling_rate, format="wav") - _audio = model.slice_inference( - temp_path, - sid, - vc_transform, - slice_db, - cluster_ratio, - auto_f0, - noise_scale, - pad_seconds, - cl_num, - lg_num, - lgr_num, - f0_predictor, - enhancer_adaptive_key, - cr_threshold, - k_step, - use_spk_mix, - second_encoding, - loudness_envelope_adjustment - ) - model.clear_empty() - os.remove(temp_path) - #构建保存文件的路径，并保存到results文件夹内 - timestamp = str(int(time.time())) - if not os.path.exists("results"): - os.makedirs("results") - output_file = os.path.join("results", sid + "_" + timestamp + ".wav") - soundfile.write(output_file, _audio, model.target_sample, format="wav") + # 未知原因Gradio上传的filepath会有一个奇怪的固定后缀，这里去掉 + truncated_basename = Path(input_audio).stem[:-6] + processed_audio = os.path.join("raw", f"{truncated_basename}.wav") + soundfile.write(processed_audio, audio, sampling_rate, format="wav") + output_file = vc_infer(output_format, sid, processed_audio, truncated_basename, vc_transform, auto_f0, cluster_ratio, slice_db, noise_scale, pad_seconds, cl_num, lg_num, lgr_num, f0_predictor, enhancer_adaptive_key, cr_threshold, k_step, use_spk_mix, second_encoding, loudness_envelope_adjustment) return "Success", output_file except Exception as e: if debug: traceback.print_exc() raise gr.Error(e) -def tts_func(_text,_rate,_voice): - #使用edge-tts把文字转成音频 - # voice = "zh-CN-XiaoyiNeural"#女性，较高音 - # voice = "zh-CN-YunxiNeural"#男性 - voice = "zh-CN-YunxiNeural"#男性 - if ( _voice == "女" ) : - voice = "zh-CN-XiaoyiNeural" - output_file = _text[0:10]+".wav" - # communicate = edge_tts.Communicate(_text, voice) - # await communicate.save(output_file) - if _rate>=0: - ratestr="+{:.0%}".format(_rate) - elif _rate<0: - ratestr="{:.0%}".format(_rate)#减号自带 - - p=subprocess.Popen("edge-tts "+ - " --text "+_text+ - " --write-media "+output_file+ - " --voice "+voice+ - " --rate="+ratestr - ,shell=True, - stdout=subprocess.PIPE, - stdin=subprocess.PIPE) - p.wait() - return output_file - def text_clear(text): return re.sub(r"[\n\,\(\) ]", "", text) -def vc_fn2(sid, input_audio, vc_transform, auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,text2tts,tts_rate,tts_voice,f0_predictor,enhancer_adaptive_key,cr_threshold): - #使用edge-tts把文字转成音频 - text2tts=text_clear(text2tts) - output_file=tts_func(text2tts,tts_rate,tts_voice) - - #调整采样率 - sr2=model.target_sample - wav, sr = librosa.load(output_file) - wav2 = librosa.resample(wav, orig_sr=sr, target_sr=sr2) - save_path2= text2tts[0:10]+"_44k"+".wav" - wavfile.write(save_path2,sr2, - (wav2 * np.iinfo(np.int16).max).astype(np.int16) - ) - - #读取音频 - sample_rate, data=gr_pu.audio_from_file(save_path2) - vc_input=(sample_rate, data) - - a,b=vc_fn(sid, vc_input, vc_transform,auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,f0_predictor,enhancer_adaptive_key,cr_threshold) - os.remove(output_file) - os.remove(save_path2) - return a,b +def vc_fn2(_text, _lang, _gender, _rate, _volume, sid, output_format, vc_transform, auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,f0_predictor,enhancer_adaptive_key,cr_threshold, k_step,use_spk_mix,second_encoding,loudness_envelope_adjustment): + global model + try: + if model is None: + return "You need to upload an model", None + if getattr(model, 'cluster_model', None) is None and model.feature_retrieval is False: + if cluster_ratio != 0: + return "You need to upload an cluster model or feature retrieval model before assigning cluster ratio!", None + _rate = f"+{int(_rate*100)}%" if _rate >= 0 else f"{int(_rate*100)}%" + _volume = f"+{int(_volume*100)}%" if _volume >= 0 else f"{int(_volume*100)}%" + if _lang == "Auto": + _gender = "Male" if _gender == "男" else "Female" + subprocess.run([r"workenv\python.exe", "edgetts/tts.py", _text, _lang, _rate, _volume, _gender]) + else: + subprocess.run([r"workenv\python.exe", "edgetts/tts.py", _text, _lang, _rate, _volume]) + target_sr = 44100 + y, sr = librosa.load("tts.wav") + resampled_y = librosa.resample(y, orig_sr=sr, target_sr=target_sr) + soundfile.write("tts.wav", resampled_y, target_sr, subtype = "PCM_16") + input_audio = "tts.wav" + #audio, _ = soundfile.read(input_audio) + output_file_path = vc_infer(output_format, sid, input_audio, "tts", vc_transform, auto_f0, cluster_ratio, slice_db, noise_scale, pad_seconds, cl_num, lg_num, lgr_num, f0_predictor, enhancer_adaptive_key, cr_threshold, k_step, use_spk_mix, second_encoding, loudness_envelope_adjustment) + os.remove("tts.wav") + return "Success", output_file_path + except Exception as e: + if debug: traceback.print_exc() + raise gr.Error(e) def model_compression(_model): if _model == "": @@ -291,6 +292,7 @@ with gr.Blocks( vc_transform = gr.Number(label="变调（整数，可以正负，半音数量，升高八度就是12）", value=0) cluster_ratio = gr.Number(label="聚类模型/特征检索混合比例，0-1之间，0即不启用聚类/特征检索。使用聚类/特征检索能提升音色相似度，但会导致咬字下降（如果使用建议0.5左右）", value=0) slice_db = gr.Number(label="切片阈值", value=-40) + output_format = gr.Radio(label="音频输出格式", choices=["wav", "flac", "mp3"], value = "wav") noise_scale = gr.Number(label="noise_scale 建议不要动，会影响音质，玄学参数", value=0.4) k_step = gr.Slider(label="浅扩散步数，只有使用了扩散模型才有效，步数越大越接近扩散模型的结果", value=100, minimum = 1, maximum = 1000) with gr.Column(): @@ -305,12 +307,15 @@ with gr.Blocks( use_spk_mix = gr.Checkbox(label = "动态声线融合", value = False, interactive = False) with gr.Tabs(): with gr.TabItem("音频转音频"): - vc_input3 = gr.Audio(label="选择音频") + vc_input3 = gr.Audio(label="选择音频", type="filepath") vc_submit = gr.Button("音频转换", variant="primary") with gr.TabItem("文字转音频"): text2tts=gr.Textbox(label="在此输入要转译的文字。注意，使用该功能建议打开F0预测，不然会很怪") - tts_rate = gr.Number(label="tts语速", value=0) - tts_voice = gr.Radio(label="性别",choices=["男","女"], value="男") + with gr.Row(): + tts_gender = gr.Radio(label = "说话人性别", choices = ["男","女"], value = "男") + tts_lang = gr.Dropdown(label = "选择语言，Auto为根据输入文字自动识别", choices=SUPPORTED_LANGUAGES, value = "Auto") + tts_rate = gr.Slider(label = "TTS语音变速（倍速相对值）", minimum = -1, maximum = 3, value = 0, step = 0.1) + tts_volume = gr.Slider(label = "TTS语音音量（相对值）", minimum = -1, maximum = 1.5, value = 0, step = 0.1) vc_submit2 = gr.Button("文字转换", variant="primary") with gr.Row(): with gr.Column(): @@ -371,8 +376,8 @@ with gr.Blocks( WebUI设置 """) debug_button = gr.Checkbox(label="Debug模式，如果向社区反馈BUG需要打开，打开后控制台可以显示具体错误提示", value=debug) - vc_submit.click(vc_fn, [sid, vc_input3, vc_transform,auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,f0_predictor,enhancer_adaptive_key,cr_threshold,k_step,use_spk_mix,second_encoding,loudness_envelope_adjustment], [vc_output1, vc_output2]) - vc_submit2.click(vc_fn2, [sid, vc_input3, vc_transform,auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,text2tts,tts_rate,tts_voice,f0_predictor,enhancer_adaptive_key,cr_threshold], [vc_output1, vc_output2]) + vc_submit.click(vc_fn, [sid, vc_input3, output_format, vc_transform,auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,f0_predictor,enhancer_adaptive_key,cr_threshold,k_step,use_spk_mix,second_encoding,loudness_envelope_adjustment], [vc_output1, vc_output2]) + vc_submit2.click(vc_fn2, [text2tts, tts_lang, tts_gender, tts_rate, tts_volume, sid, output_format, vc_transform,auto_f0,cluster_ratio, slice_db, noise_scale,pad_seconds,cl_num,lg_num,lgr_num,f0_predictor,enhancer_adaptive_key,cr_threshold,k_step,use_spk_mix,second_encoding,loudness_envelope_adjustment], [vc_output1, vc_output2]) debug_button.change(debug_change,[],[]) model_load_button.click(modelAnalysis,[model_path,config_path,cluster_model_path,device,enhance,diff_model_path,diff_config_path,only_diffusion,use_spk_mix],[sid,sid_output]) model_unload_button.click(modelUnload,[],[sid,sid_output]) From 9f13c8e43c19ca68015c007590ff4f22df79feff Mon Sep 17 00:00:00 2001 From: umoubuton <127150330+umoufuton@users.noreply.github.com> Date: Wed, 5 Jul 2023 21:55:43 +0800 Subject: [PATCH 2/4] Update requirements --- requirements.txt | 1 + requirements_onnx_encoder.txt | 4 ++-- requirements_win.txt | 1 + 3 files changed, 4 insertions(+), 2 deletions(-) diff --git a/requirements.txt b/requirements.txt index d09b883..5b6b5f8 100644 --- a/requirements.txt +++ b/requirements.txt @@ -21,6 +21,7 @@ tensorboard tensorboardX transformers edge_tts +langdetect pyyaml pynvml faiss-cpu diff --git a/requirements_onnx_encoder.txt b/requirements_onnx_encoder.txt index 4263943..a624622 100644 --- a/requirements_onnx_encoder.txt +++ b/requirements_onnx_encoder.txt @@ -19,9 +19,9 @@ librosa==0.9.1 tensorboard tensorboardX edge_tts +langdetect pyyaml pynvml transformers ffmpeg-python -ffmpeg -faiss \ No newline at end of file +faiss-cpu \ No newline at end of file diff --git a/requirements_win.txt b/requirements_win.txt index e35a80e..d7f701e 100644 --- a/requirements_win.txt +++ b/requirements_win.txt @@ -24,6 +24,7 @@ onnxoptimizer tensorboardX transformers edge_tts +langdetect pyyaml pynvml faiss-cpu From d4f6b7361f3b6db9b5148bb812f323f78ad43975 Mon Sep 17 00:00:00 2001 From: umoubuton <127150330+umoufuton@users.noreply.github.com> Date: Wed, 5 Jul 2023 22:01:03 +0800 Subject: [PATCH 3/4] debug --- webUI.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/webUI.py b/webUI.py index c4ded2d..b0b9052 100644 --- a/webUI.py +++ b/webUI.py @@ -214,9 +214,9 @@ def vc_fn2(_text, _lang, _gender, _rate, _volume, sid, output_format, vc_transfo _volume = f"+{int(_volume*100)}%" if _volume >= 0 else f"{int(_volume*100)}%" if _lang == "Auto": _gender = "Male" if _gender == "男" else "Female" - subprocess.run([r"workenv\python.exe", "edgetts/tts.py", _text, _lang, _rate, _volume, _gender]) + subprocess.run([r"python", "edgetts/tts.py", _text, _lang, _rate, _volume, _gender]) else: - subprocess.run([r"workenv\python.exe", "edgetts/tts.py", _text, _lang, _rate, _volume]) + subprocess.run([r"python", "edgetts/tts.py", _text, _lang, _rate, _volume]) target_sr = 44100 y, sr = librosa.load("tts.wav") resampled_y = librosa.resample(y, orig_sr=sr, target_sr=target_sr) From c887fd3cfb4a92f27e27a961400b0a9751df984f Mon Sep 17 00:00:00 2001 From: umoubuton <127150330+umoufuton@users.noreply.github.com> Date: Wed, 5 Jul 2023 22:21:25 +0800 Subject: [PATCH 4/4] debug --- edgetts/tts.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/edgetts/tts.py b/edgetts/tts.py index 03fd5bf..252ad28 100644 --- a/edgetts/tts.py +++ b/edgetts/tts.py @@ -10,8 +10,6 @@ DetectorFactory.seed = 0 TEXT = sys.argv[1] LANG = detect(TEXT) if sys.argv[2] == "Auto" else sys.argv[2] -if LANG == "zh-cn" or LANG == "zh-tw": - LOCALE = LANG[:-2] + LANG[-2:].upper() RATE = sys.argv[3] VOLUME = sys.argv[4] GENDER = sys.argv[5] if len(sys.argv) == 6 else None @@ -23,7 +21,9 @@ print(f"Text: {TEXT}, Language: {LANG}, Gender: {GENDER}, Rate: {RATE}, Volume: async def _main() -> None: voices = await VoicesManager.create() if not GENDER is None: + # From "zh-cn" to "zh-CN" etc. if LANG == "zh-cn" or LANG == "zh-tw": + LOCALE = LANG[:-2] + LANG[-2:].upper() voice = voices.find(Gender=GENDER, Locale=LOCALE) else: voice = voices.find(Gender=GENDER, Language=LANG) @@ -44,4 +44,4 @@ if __name__ == "__main__": try: loop.run_until_complete(_main()) finally: - loop.close() \ No newline at end of file + loop.close()