[ADD] edge tts

app.py

@@ -6,6 +6,8 @@ import hashlib
 from utils.model import model_downloader, get_model
 import requests
 import json
+from tts.constants import VOICE_METHODS, BARK_VOICES, EDGE_VOICES
+from tts.conversion import tts_infer
 
 api_url = "https://rvc-models-api.onrender.com/uploadfile/"
 
@@ -67,6 +69,7 @@ def infer(model, f0_method, audio_file):
     else:
         return
     
+
 def post_model(name, model_url, version, creator):
     modelname = model_downloader(model_url, zips_folder, unzips_folder)
     model_files = get_model(unzips_folder, modelname)
@@ -116,6 +119,12 @@ def search_model(name):
                 result.append(f"**Nombre del modelo: {model_name}**</br>{model_url}</br>")
                 yield "</br>".join(result)
             cont += 1
+            
+def update_tts_methods_voice(select_value):
+    if select_value == "Edge-tts":
+        return gr.update(choices=EDGE_VOICES)
+    elif select_value == "Bark-tts":
+        return gr.update(choices=BARK_VOICES)
 
 with gr.Blocks() as app:
     gr.HTML("<h1> Simple RVC Inference - by Juuxn 💻 </h1>")
@@ -134,6 +143,30 @@ with gr.Blocks() as app:
         btn = gr.Button(value="Convertir")
         btn.click(infer, inputs=[model_url, f0_method, audio_path], outputs=[vc_output1, vc_output2])
         
+    with gr.TabItem("TTS"):
+        with gr.Row():
+            tts_text = gr.Textbox(
+                label="Texto:",
+                placeholder="Texto que deseas convertir a voz...",
+                lines=6,
+            )
+
+        with gr.Column():
+            with gr.Row():
+                tts_model_url = gr.Textbox(placeholder="https://huggingface.co/AIVER-SE/BillieEilish/resolve/main/BillieEilish.zip", label="Url del modelo RVC", show_label=True)
+                
+            with gr.Column():
+                tts_method = gr.Dropdown(choices=VOICE_METHODS, value="Edge-tts", label="Método TTS:", visible=False)
+                tts_model = gr.Dropdown(choices=EDGE_VOICES, label="Modelo TTS:", visible=True, interactive=True)
+                tts_method.change(fn=update_tts_methods_voice, inputs=[tts_method], outputs=[tts_model])
+                    
+            with gr.Row():
+                tts_vc_output1 = gr.Textbox(label="Salida")
+                tts_vc_output2 = gr.Audio(label="Audio de salida")   
+            
+        tts_btn = gr.Button(value="Convertir")
+        tts_btn.click(fn=tts_infer, inputs=[tts_text, tts_model_url, tts_method, tts_model], outputs=[tts_vc_output1, tts_vc_output2])
+        
     with gr.Tab("Recursos"):
         gr.HTML("<h4>Buscar modelos</h4>")
         search_name = gr.Textbox(placeholder="Billie Eillish (RVC v2 - 100 epoch)", label="Nombre", show_label=True)
@@ -156,5 +189,58 @@ with gr.Blocks() as app:
             
         btn_post_model = gr.Button(value="Publicar")
         btn_post_model.click(fn=post_model, inputs=[post_name, post_model_url, post_version, post_creator], outputs=[post_output])
+                
+
+        #     with gr.Column():
+        #         model_voice_path07 = gr.Dropdown(
+        #             label=i18n("RVC Model:"),
+        #             choices=sorted(names),
+        #             value=default_weight,
+        #         )
+        #         best_match_index_path1, _ = match_index(
+        #             model_voice_path07.value
+        #         )
+
+        #         file_index2_07 = gr.Dropdown(
+        #             label=i18n("Select the .index file:"),
+        #             choices=get_indexes(),
+        #             value=best_match_index_path1,
+        #             interactive=True,
+        #             allow_custom_value=True,
+        #         )
+        # with gr.Row():
+        #     refresh_button_ = gr.Button(i18n("Refresh"), variant="primary")
+        #     refresh_button_.click(
+        #         fn=change_choices2,
+        #         inputs=[],
+        #         outputs=[model_voice_path07, file_index2_07],
+        #     )
+        # with gr.Row():
+        #     original_ttsvoice = gr.Audio(label=i18n("Audio TTS:"))
+        #     ttsvoice = gr.Audio(label=i18n("Audio RVC:"))
+
+        # with gr.Row():
+        #     button_test = gr.Button(i18n("Convert"), variant="primary")
+
+        # button_test.click(
+        #     tts.use_tts,
+        #     inputs=[
+        #         text_test,
+        #         tts_test,
+        #         model_voice_path07,
+        #         file_index2_07,
+        #         # transpose_test,
+        #         vc_transform0,
+        #         f0method8,
+        #         index_rate1,
+        #         crepe_hop_length,
+        #         f0_autotune,
+        #         ttsmethod_test,
+        #     ],
+        #     outputs=[ttsvoice, original_ttsvoice],
+        # )
+
+    
     
-    app.queue(concurrency_count=511, max_size=1022).launch(share=True)
+    app.queue(concurrency_count=511, max_size=1022).launch()
+    #share=True

requirements.txt

@@ -1,3 +1,4 @@
+[email protected]:huggingface/transformers.git@main#egg=transformers
 absl-py==1.4.0
 aiofiles==23.1.0
 aiohttp==3.8.3
@@ -44,7 +45,7 @@ fsspec==2022.11.0
 future==0.18.2
 google-auth==2.16.2
 google-auth-oauthlib==1.0.0
-gradio==3.34.0
+gradio
 gradio_client==0.2.10
 grpcio==1.51.3
 gspread==5.10.0
@@ -166,5 +167,7 @@ yarl==1.8.2
 zipp==3.15.0
 firebase
 firebase_admin
+nltk
 gdown
-validators
+validators
+git+https://github.com/suno-ai/bark.git

tts/constants.py

@@ -0,0 +1,438 @@
+VOICE_METHODS = ["Edge-tts", "Bark-tts"]
+
+BARK_VOICES = [
+    "v2/en_speaker_0-Male",
+    "v2/en_speaker_1-Male",
+    "v2/en_speaker_2-Male",
+    "v2/en_speaker_3-Male",
+    "v2/en_speaker_4-Male",
+    "v2/en_speaker_5-Male",
+    "v2/en_speaker_6-Male",
+    "v2/en_speaker_7-Male",
+    "v2/en_speaker_8-Male",
+    "v2/en_speaker_9-Female",
+    "v2/zh_speaker_0-Male",
+    "v2/zh_speaker_1-Male",
+    "v2/zh_speaker_2-Male",
+    "v2/zh_speaker_3-Male",
+    "v2/zh_speaker_4-Female",
+    "v2/zh_speaker_5-Male",
+    "v2/zh_speaker_6-Female",
+    "v2/zh_speaker_7-Female",
+    "v2/zh_speaker_8-Male",
+    "v2/zh_speaker_9-Female",
+    "v2/fr_speaker_0-Male",
+    "v2/fr_speaker_1-Female",
+    "v2/fr_speaker_2-Female",
+    "v2/fr_speaker_3-Male",
+    "v2/fr_speaker_4-Male",
+    "v2/fr_speaker_5-Female",
+    "v2/fr_speaker_6-Male",
+    "v2/fr_speaker_7-Male",
+    "v2/fr_speaker_8-Male",
+    "v2/fr_speaker_9-Male",
+    "v2/de_speaker_0-Male",
+    "v2/de_speaker_1-Male",
+    "v2/de_speaker_2-Male",
+    "v2/de_speaker_3-Female",
+    "v2/de_speaker_4-Male",
+    "v2/de_speaker_5-Male",
+    "v2/de_speaker_6-Male",
+    "v2/de_speaker_7-Male",
+    "v2/de_speaker_8-Female",
+    "v2/de_speaker_9-Male",
+    "v2/hi_speaker_0-Female",
+    "v2/hi_speaker_1-Female",
+    "v2/hi_speaker_2-Male",
+    "v2/hi_speaker_3-Female",
+    "v2/hi_speaker_4-Female",
+    "v2/hi_speaker_5-Male",
+    "v2/hi_speaker_6-Male",
+    "v2/hi_speaker_7-Male",
+    "v2/hi_speaker_8-Male",
+    "v2/hi_speaker_9-Female",
+    "v2/it_speaker_0-Male",
+    "v2/it_speaker_1-Male",
+    "v2/it_speaker_2-Female",
+    "v2/it_speaker_3-Male",
+    "v2/it_speaker_4-Male",
+    "v2/it_speaker_5-Male",
+    "v2/it_speaker_6-Male",
+    "v2/it_speaker_7-Female",
+    "v2/it_speaker_8-Male",
+    "v2/it_speaker_9-Female",
+    "v2/ja_speaker_0-Female",
+    "v2/ja_speaker_1-Female",
+    "v2/ja_speaker_2-Male",
+    "v2/ja_speaker_3-Female",
+    "v2/ja_speaker_4-Female",
+    "v2/ja_speaker_5-Female",
+    "v2/ja_speaker_6-Male",
+    "v2/ja_speaker_7-Female",
+    "v2/ja_speaker_8-Female",
+    "v2/ja_speaker_9-Female",
+    "v2/ko_speaker_0-Female",
+    "v2/ko_speaker_1-Male",
+    "v2/ko_speaker_2-Male",
+    "v2/ko_speaker_3-Male",
+    "v2/ko_speaker_4-Male",
+    "v2/ko_speaker_5-Male",
+    "v2/ko_speaker_6-Male",
+    "v2/ko_speaker_7-Male",
+    "v2/ko_speaker_8-Male",
+    "v2/ko_speaker_9-Male",
+    "v2/pl_speaker_0-Male",
+    "v2/pl_speaker_1-Male",
+    "v2/pl_speaker_2-Male",
+    "v2/pl_speaker_3-Male",
+    "v2/pl_speaker_4-Female",
+    "v2/pl_speaker_5-Male",
+    "v2/pl_speaker_6-Female",
+    "v2/pl_speaker_7-Male",
+    "v2/pl_speaker_8-Male",
+    "v2/pl_speaker_9-Female",
+    "v2/pt_speaker_0-Male",
+    "v2/pt_speaker_1-Male",
+    "v2/pt_speaker_2-Male",
+    "v2/pt_speaker_3-Male",
+    "v2/pt_speaker_4-Male",
+    "v2/pt_speaker_5-Male",
+    "v2/pt_speaker_6-Male",
+    "v2/pt_speaker_7-Male",
+    "v2/pt_speaker_8-Male",
+    "v2/pt_speaker_9-Male",
+    "v2/ru_speaker_0-Male",
+    "v2/ru_speaker_1-Male",
+    "v2/ru_speaker_2-Male",
+    "v2/ru_speaker_3-Male",
+    "v2/ru_speaker_4-Male",
+    "v2/ru_speaker_5-Female",
+    "v2/ru_speaker_6-Female",
+    "v2/ru_speaker_7-Male",
+    "v2/ru_speaker_8-Male",
+    "v2/ru_speaker_9-Female",
+    "v2/es_speaker_0-Male",
+    "v2/es_speaker_1-Male",
+    "v2/es_speaker_2-Male",
+    "v2/es_speaker_3-Male",
+    "v2/es_speaker_4-Male",
+    "v2/es_speaker_5-Male",
+    "v2/es_speaker_6-Male",
+    "v2/es_speaker_7-Male",
+    "v2/es_speaker_8-Female",
+    "v2/es_speaker_9-Female",
+    "v2/tr_speaker_0-Male",
+    "v2/tr_speaker_1-Male",
+    "v2/tr_speaker_2-Male",
+    "v2/tr_speaker_3-Male",
+    "v2/tr_speaker_4-Female",
+    "v2/tr_speaker_5-Female",
+    "v2/tr_speaker_6-Male",
+    "v2/tr_speaker_7-Male",
+    "v2/tr_speaker_8-Male",
+    "v2/tr_speaker_9-Male",
+]
+
+EDGE_VOICES = [
+    "af-ZA-AdriNeural-Female",
+    "af-ZA-WillemNeural-Male",
+    "sq-AL-AnilaNeural-Female",
+    "sq-AL-IlirNeural-Male",
+    "am-ET-AmehaNeural-Male",
+    "am-ET-MekdesNeural-Female",
+    "ar-DZ-AminaNeural-Female",
+    "ar-DZ-IsmaelNeural-Male",
+    "ar-BH-AliNeural-Male",
+    "ar-BH-LailaNeural-Female",
+    "ar-EG-SalmaNeural-Female",
+    "ar-EG-ShakirNeural-Male",
+    "ar-IQ-BasselNeural-Male",
+    "ar-IQ-RanaNeural-Female",
+    "ar-JO-SanaNeural-Female",
+    "ar-JO-TaimNeural-Male",
+    "ar-KW-FahedNeural-Male",
+    "ar-KW-NouraNeural-Female",
+    "ar-LB-LaylaNeural-Female",
+    "ar-LB-RamiNeural-Male",
+    "ar-LY-ImanNeural-Female",
+    "ar-LY-OmarNeural-Male",
+    "ar-MA-JamalNeural-Male",
+    "ar-MA-MounaNeural-Female",
+    "ar-OM-AbdullahNeural-Male",
+    "ar-OM-AyshaNeural-Female",
+    "ar-QA-AmalNeural-Female",
+    "ar-QA-MoazNeural-Male",
+    "ar-SA-HamedNeural-Male",
+    "ar-SA-ZariyahNeural-Female",
+    "ar-SY-AmanyNeural-Female",
+    "ar-SY-LaithNeural-Male",
+    "ar-TN-HediNeural-Male",
+    "ar-TN-ReemNeural-Female",
+    "ar-AE-FatimaNeural-Female",
+    "ar-AE-HamdanNeural-Male",
+    "ar-YE-MaryamNeural-Female",
+    "ar-YE-SalehNeural-Male",
+    "az-AZ-BabekNeural-Male",
+    "az-AZ-BanuNeural-Female",
+    "bn-BD-NabanitaNeural-Female",
+    "bn-BD-PradeepNeural-Male",
+    "bn-IN-BashkarNeural-Male",
+    "bn-IN-TanishaaNeural-Female",
+    "bs-BA-GoranNeural-Male",
+    "bs-BA-VesnaNeural-Female",
+    "bg-BG-BorislavNeural-Male",
+    "bg-BG-KalinaNeural-Female",
+    "my-MM-NilarNeural-Female",
+    "my-MM-ThihaNeural-Male",
+    "ca-ES-EnricNeural-Male",
+    "ca-ES-JoanaNeural-Female",
+    "zh-HK-HiuGaaiNeural-Female",
+    "zh-HK-HiuMaanNeural-Female",
+    "zh-HK-WanLungNeural-Male",
+    "zh-CN-XiaoxiaoNeural-Female",
+    "zh-CN-XiaoyiNeural-Female",
+    "zh-CN-YunjianNeural-Male",
+    "zh-CN-YunxiNeural-Male",
+    "zh-CN-YunxiaNeural-Male",
+    "zh-CN-YunyangNeural-Male",
+    "zh-CN-liaoning-XiaobeiNeural-Female",
+    "zh-TW-HsiaoChenNeural-Female",
+    "zh-TW-YunJheNeural-Male",
+    "zh-TW-HsiaoYuNeural-Female",
+    "zh-CN-shaanxi-XiaoniNeural-Female",
+    "hr-HR-GabrijelaNeural-Female",
+    "hr-HR-SreckoNeural-Male",
+    "cs-CZ-AntoninNeural-Male",
+    "cs-CZ-VlastaNeural-Female",
+    "da-DK-ChristelNeural-Female",
+    "da-DK-JeppeNeural-Male",
+    "nl-BE-ArnaudNeural-Male",
+    "nl-BE-DenaNeural-Female",
+    "nl-NL-ColetteNeural-Female",
+    "nl-NL-FennaNeural-Female",
+    "nl-NL-MaartenNeural-Male",
+    "en-AU-NatashaNeural-Female",
+    "en-AU-WilliamNeural-Male",
+    "en-CA-ClaraNeural-Female",
+    "en-CA-LiamNeural-Male",
+    "en-HK-SamNeural-Male",
+    "en-HK-YanNeural-Female",
+    "en-IN-NeerjaExpressiveNeural-Female",
+    "en-IN-NeerjaNeural-Female",
+    "en-IN-PrabhatNeural-Male",
+    "en-IE-ConnorNeural-Male",
+    "en-IE-EmilyNeural-Female",
+    "en-KE-AsiliaNeural-Female",
+    "en-KE-ChilembaNeural-Male",
+    "en-NZ-MitchellNeural-Male",
+    "en-NZ-MollyNeural-Female",
+    "en-NG-AbeoNeural-Male",
+    "en-NG-EzinneNeural-Female",
+    "en-PH-JamesNeural-Male",
+    "en-PH-RosaNeural-Female",
+    "en-SG-LunaNeural-Female",
+    "en-SG-WayneNeural-Male",
+    "en-ZA-LeahNeural-Female",
+    "en-ZA-LukeNeural-Male",
+    "en-TZ-ElimuNeural-Male",
+    "en-TZ-ImaniNeural-Female",
+    "en-GB-LibbyNeural-Female",
+    "en-GB-MaisieNeural-Female",
+    "en-GB-RyanNeural-Male",
+    "en-GB-SoniaNeural-Female",
+    "en-GB-ThomasNeural-Male",
+    "en-US-AriaNeural-Female",
+    "en-US-AnaNeural-Female",
+    "en-US-ChristopherNeural-Male",
+    "en-US-EricNeural-Male",
+    "en-US-GuyNeural-Male",
+    "en-US-JennyNeural-Female",
+    "en-US-MichelleNeural-Female",
+    "en-US-RogerNeural-Male",
+    "en-US-SteffanNeural-Male",
+    "et-EE-AnuNeural-Female",
+    "et-EE-KertNeural-Male",
+    "fil-PH-AngeloNeural-Male",
+    "fil-PH-BlessicaNeural-Female",
+    "fi-FI-HarriNeural-Male",
+    "fi-FI-NooraNeural-Female",
+    "fr-BE-CharlineNeural-Female",
+    "fr-BE-GerardNeural-Male",
+    "fr-CA-AntoineNeural-Male",
+    "fr-CA-JeanNeural-Male",
+    "fr-CA-SylvieNeural-Female",
+    "fr-FR-DeniseNeural-Female",
+    "fr-FR-EloiseNeural-Female",
+    "fr-FR-HenriNeural-Male",
+    "fr-CH-ArianeNeural-Female",
+    "fr-CH-FabriceNeural-Male",
+    "gl-ES-RoiNeural-Male",
+    "gl-ES-SabelaNeural-Female",
+    "ka-GE-EkaNeural-Female",
+    "ka-GE-GiorgiNeural-Male",
+    "de-AT-IngridNeural-Female",
+    "de-AT-JonasNeural-Male",
+    "de-DE-AmalaNeural-Female",
+    "de-DE-ConradNeural-Male",
+    "de-DE-KatjaNeural-Female",
+    "de-DE-KillianNeural-Male",
+    "de-CH-JanNeural-Male",
+    "de-CH-LeniNeural-Female",
+    "el-GR-AthinaNeural-Female",
+    "el-GR-NestorasNeural-Male",
+    "gu-IN-DhwaniNeural-Female",
+    "gu-IN-NiranjanNeural-Male",
+    "he-IL-AvriNeural-Male",
+    "he-IL-HilaNeural-Female",
+    "hi-IN-MadhurNeural-Male",
+    "hi-IN-SwaraNeural-Female",
+    "hu-HU-NoemiNeural-Female",
+    "hu-HU-TamasNeural-Male",
+    "is-IS-GudrunNeural-Female",
+    "is-IS-GunnarNeural-Male",
+    "id-ID-ArdiNeural-Male",
+    "id-ID-GadisNeural-Female",
+    "ga-IE-ColmNeural-Male",
+    "ga-IE-OrlaNeural-Female",
+    "it-IT-DiegoNeural-Male",
+    "it-IT-ElsaNeural-Female",
+    "it-IT-IsabellaNeural-Female",
+    "ja-JP-KeitaNeural-Male",
+    "ja-JP-NanamiNeural-Female",
+    "jv-ID-DimasNeural-Male",
+    "jv-ID-SitiNeural-Female",
+    "kn-IN-GaganNeural-Male",
+    "kn-IN-SapnaNeural-Female",
+    "kk-KZ-AigulNeural-Female",
+    "kk-KZ-DauletNeural-Male",
+    "km-KH-PisethNeural-Male",
+    "km-KH-SreymomNeural-Female",
+    "ko-KR-InJoonNeural-Male",
+    "ko-KR-SunHiNeural-Female",
+    "lo-LA-ChanthavongNeural-Male",
+    "lo-LA-KeomanyNeural-Female",
+    "lv-LV-EveritaNeural-Female",
+    "lv-LV-NilsNeural-Male",
+    "lt-LT-LeonasNeural-Male",
+    "lt-LT-OnaNeural-Female",
+    "mk-MK-AleksandarNeural-Male",
+    "mk-MK-MarijaNeural-Female",
+    "ms-MY-OsmanNeural-Male",
+    "ms-MY-YasminNeural-Female",
+    "ml-IN-MidhunNeural-Male",
+    "ml-IN-SobhanaNeural-Female",
+    "mt-MT-GraceNeural-Female",
+    "mt-MT-JosephNeural-Male",
+    "mr-IN-AarohiNeural-Female",
+    "mr-IN-ManoharNeural-Male",
+    "mn-MN-BataaNeural-Male",
+    "mn-MN-YesuiNeural-Female",
+    "ne-NP-HemkalaNeural-Female",
+    "ne-NP-SagarNeural-Male",
+    "nb-NO-FinnNeural-Male",
+    "nb-NO-PernilleNeural-Female",
+    "ps-AF-GulNawazNeural-Male",
+    "ps-AF-LatifaNeural-Female",
+    "fa-IR-DilaraNeural-Female",
+    "fa-IR-FaridNeural-Male",
+    "pl-PL-MarekNeural-Male",
+    "pl-PL-ZofiaNeural-Female",
+    "pt-BR-AntonioNeural-Male",
+    "pt-BR-FranciscaNeural-Female",
+    "pt-PT-DuarteNeural-Male",
+    "pt-PT-RaquelNeural-Female",
+    "ro-RO-AlinaNeural-Female",
+    "ro-RO-EmilNeural-Male",
+    "ru-RU-DmitryNeural-Male",
+    "ru-RU-SvetlanaNeural-Female",
+    "sr-RS-NicholasNeural-Male",
+    "sr-RS-SophieNeural-Female",
+    "si-LK-SameeraNeural-Male",
+    "si-LK-ThiliniNeural-Female",
+    "sk-SK-LukasNeural-Male",
+    "sk-SK-ViktoriaNeural-Female",
+    "sl-SI-PetraNeural-Female",
+    "sl-SI-RokNeural-Male",
+    "so-SO-MuuseNeural-Male",
+    "so-SO-UbaxNeural-Female",
+    "es-AR-ElenaNeural-Female",
+    "es-AR-TomasNeural-Male",
+    "es-BO-MarceloNeural-Male",
+    "es-BO-SofiaNeural-Female",
+    "es-CL-CatalinaNeural-Female",
+    "es-CL-LorenzoNeural-Male",
+    "es-CO-GonzaloNeural-Male",
+    "es-CO-SalomeNeural-Female",
+    "es-CR-JuanNeural-Male",
+    "es-CR-MariaNeural-Female",
+    "es-CU-BelkysNeural-Female",
+    "es-CU-ManuelNeural-Male",
+    "es-DO-EmilioNeural-Male",
+    "es-DO-RamonaNeural-Female",
+    "es-EC-AndreaNeural-Female",
+    "es-EC-LuisNeural-Male",
+    "es-SV-LorenaNeural-Female",
+    "es-SV-RodrigoNeural-Male",
+    "es-GQ-JavierNeural-Male",
+    "es-GQ-TeresaNeural-Female",
+    "es-GT-AndresNeural-Male",
+    "es-GT-MartaNeural-Female",
+    "es-HN-CarlosNeural-Male",
+    "es-HN-KarlaNeural-Female",
+    "es-MX-DaliaNeural-Female",
+    "es-MX-JorgeNeural-Male",
+    "es-NI-FedericoNeural-Male",
+    "es-NI-YolandaNeural-Female",
+    "es-PA-MargaritaNeural-Female",
+    "es-PA-RobertoNeural-Male",
+    "es-PY-MarioNeural-Male",
+    "es-PY-TaniaNeural-Female",
+    "es-PE-AlexNeural-Male",
+    "es-PE-CamilaNeural-Female",
+    "es-PR-KarinaNeural-Female",
+    "es-PR-VictorNeural-Male",
+    "es-ES-AlvaroNeural-Male",
+    "es-ES-ElviraNeural-Female",
+    "es-US-AlonsoNeural-Male",
+    "es-US-PalomaNeural-Female",
+    "es-UY-MateoNeural-Male",
+    "es-UY-ValentinaNeural-Female",
+    "es-VE-PaolaNeural-Female",
+    "es-VE-SebastianNeural-Male",
+    "su-ID-JajangNeural-Male",
+    "su-ID-TutiNeural-Female",
+    "sw-KE-RafikiNeural-Male",
+    "sw-KE-ZuriNeural-Female",
+    "sw-TZ-DaudiNeural-Male",
+    "sw-TZ-RehemaNeural-Female",
+    "sv-SE-MattiasNeural-Male",
+    "sv-SE-SofieNeural-Female",
+    "ta-IN-PallaviNeural-Female",
+    "ta-IN-ValluvarNeural-Male",
+    "ta-MY-KaniNeural-Female",
+    "ta-MY-SuryaNeural-Male",
+    "ta-SG-AnbuNeural-Male",
+    "ta-SG-VenbaNeural-Female",
+    "ta-LK-KumarNeural-Male",
+    "ta-LK-SaranyaNeural-Female",
+    "te-IN-MohanNeural-Male",
+    "te-IN-ShrutiNeural-Female",
+    "th-TH-NiwatNeural-Male",
+    "th-TH-PremwadeeNeural-Female",
+    "tr-TR-AhmetNeural-Male",
+    "tr-TR-EmelNeural-Female",
+    "uk-UA-OstapNeural-Male",
+    "uk-UA-PolinaNeural-Female",
+    "ur-IN-GulNeural-Female",
+    "ur-IN-SalmanNeural-Male",
+    "ur-PK-AsadNeural-Male",
+    "ur-PK-UzmaNeural-Female",
+    "uz-UZ-MadinaNeural-Female",
+    "uz-UZ-SardorNeural-Male",
+    "vi-VN-HoaiMyNeural-Female",
+    "vi-VN-NamMinhNeural-Male",
+    "cy-GB-AledNeural-Male",
+    "cy-GB-NiaNeural-Female",
+    "zu-ZA-ThandoNeural-Female",
+    "zu-ZA-ThembaNeural-Male",
+]

tts/conversion.py

@@ -0,0 +1,138 @@
+import os
+import uuid
+import numpy as np
+import torch
+import soundfile as sf
+from gtts import gTTS
+import edge_tts
+from inference import Inference
+import asyncio
+#git+https://github.com/suno-ai/bark.git
+# from transformers import AutoProcessor, BarkModel
+# import nltk
+# from nltk.tokenize import sent_tokenize
+# from bark import SAMPLE_RATE
+
+# now_dir = os.getcwd()
+
+def cast_to_device(tensor, device):
+    try:
+        return tensor.to(device)
+    except Exception as e:
+        print(e)
+        return tensor
+
+# Buscar la forma de evitar descargar el archivo de 4gb cada vez que crea una instancia
+# def _bark_conversion_(text, voice_preset):
+#     os.makedirs(os.path.join(now_dir, "tts"), exist_ok=True)
+
+#     device = "cuda:0" if torch.cuda.is_available() else "cpu"
+#     dtype = torch.float32 if "cpu" in device else torch.float16
+#     bark_processor = AutoProcessor.from_pretrained(
+#         "suno/bark",
+#         cache_dir=os.path.join(now_dir, "tts", "suno/bark"),
+#         torch_dtype=dtype,
+#     )
+#     bark_model = BarkModel.from_pretrained(
+#         "suno/bark",
+#         cache_dir=os.path.join(now_dir, "tts", "suno/bark"),
+#         torch_dtype=dtype,
+#     ).to(device)
+#     # bark_model.enable_cpu_offload()
+#     inputs = bark_processor(text=[text], return_tensors="pt", voice_preset=voice_preset)
+#     tensor_dict = {
+#         k: cast_to_device(v, device) if hasattr(v, "to") else v
+#         for k, v in inputs.items()
+#     }
+#     speech_values = bark_model.generate(**tensor_dict, do_sample=True)
+#     sampling_rate = bark_model.generation_config.sample_rate
+#     speech = speech_values.cpu().numpy().squeeze()
+#     return speech, sampling_rate
+
+
+def tts_infer(tts_text, model_url, tts_method, tts_model):
+    print("*****************")
+    print(tts_text)
+    print(model_url)
+    if not tts_text:
+        return 'Primero escribe el texto que quieres convertir.', None
+    if not tts_model:
+        return 'Selecciona un modelo TTS antes de convertir.', None
+    if not model_url:
+        return 'Escribe la url de modelo que quieres usar antes de convertir.', None
+        
+    f0_method = "harvest" 
+    output_folder = "audios"
+    os.makedirs(output_folder, exist_ok=True)
+    converted_tts_filename = os.path.join(output_folder, f"tts_out_{uuid.uuid4()}.wav")
+    success = False
+
+    if len(tts_text) > 60:
+        tts_text = tts_text[:60]
+        print("DEMO; limit to 60 characters")
+
+    language = tts_model[:2]
+    if tts_method == "Edge-tts":
+        try:
+            asyncio.run(
+                edge_tts.Communicate(
+                    tts_text, "-".join(tts_model.split("-")[:-1])
+                ).save(converted_tts_filename)
+            )
+            success = True
+        except Exception as e:
+            print("ERROR", e)
+            try:
+                tts = gTTS(tts_text, lang=language)
+                tts.save(converted_tts_filename)
+                print(
+                    f"No audio was received. Please change the tts voice for {tts_model}. USING gTTS."
+                )
+                success = True
+            except:
+                tts = gTTS("a", lang=language)
+                tts.save(converted_tts_filename)
+                print("Error: Audio will be replaced.")
+                success = False
+                
+    # elif tts_method == "Bark-tts":
+    #     try:
+    #         script = tts_text.replace("\n", " ").strip()
+    #         sentences = sent_tokenize(script)
+    #         silence = np.zeros(int(0.25 * SAMPLE_RATE))
+    #         pieces = []
+    #         for sentence in sentences:
+    #             audio_array, _ = _bark_conversion_(sentence, tts_model.split("-")[0])
+    #             pieces += [audio_array, silence.copy()]
+
+    #         sf.write(
+    #             file=converted_tts_filename, samplerate=SAMPLE_RATE, data=np.concatenate(pieces)
+    #         )
+            
+    #     except Exception as e:
+    #         print(f"{e}")
+    #         return None, None
+    
+    if success:
+        inference = Inference(
+            model_name=model_url,
+            f0_method=f0_method,
+            source_audio_path=converted_tts_filename,
+            output_file_name=os.path.join("./audio-outputs", os.path.basename(converted_tts_filename)),
+        )
+        output = inference.run()
+        if os.path.exists(converted_tts_filename):
+            os.remove(converted_tts_filename)
+        
+        if os.path.exists(os.path.join("weights", inference.model_name)):
+            os.remove(os.path.join("weights", inference.model_name))
+            
+        if 'success' in output and output['success']:
+            return output, output['file']
+        else:
+            return output, None
+    else:
+        return "Ocurrió un error durante la conversión", None
+    
+    
+