Release 6.0.0 model

robinhad · Sep 17, 2023 · 316ae6b · 316ae6b
1 parent dc248e1
commit 316ae6b
Show file tree

Hide file tree

Showing 7 changed files with 166 additions and 162 deletions.
diff --git a/.gitignore b/.gitignore
@@ -135,6 +135,7 @@ dmypy.json
 *.pth.tar
 *.pth
 *.ark
+*.npz
 
 # gradio
 gradio_queue.db

diff --git a/README.md b/README.md
@@ -38,27 +38,30 @@ If you like my work, please support ❤️ -> [https://send.monobank.ua/jar/48iH
 You're welcome to join UA Speech Recognition and Synthesis community: [Telegram https://t.me/speech_recognition_uk](https://t.me/speech_recognition_uk)
 # Examples 🤖
 
-`Tetiana (female)`:
+`Oleksa (male)`:
 
-https://user-images.githubusercontent.com/5759207/224504324-d8236cad-7302-4dfd-9696-7a42b9f05fce.mp4
+https://github.com/robinhad/ukrainian-tts/assets/5759207/ace842ef-06d0-4b1f-ad49-5fda92999dbb
 
 
 <details>
   <summary>More voices 📢🤖</summary>
 
-`Dmytro (male)`:
+`Tetiana (female)`:
 
-https://user-images.githubusercontent.com/5759207/224504354-f84f74d3-fa46-497c-9604-4b63ba45989f.mp4
+https://github.com/robinhad/ukrainian-tts/assets/5759207/a6ecacf6-62ae-4fc5-b6d5-41e6cdd3d992
 
+`Dmytro (male)`:
 
-`Lada (female)`:
+https://github.com/robinhad/ukrainian-tts/assets/5759207/67d3dac9-6626-40ef-98e5-ec194096bbe0
 
-https://user-images.githubusercontent.com/5759207/224504360-ec198ac2-647c-4238-99ef-b6f074d633fd.mp4
+`Lada (female)`:
 
+https://github.com/robinhad/ukrainian-tts/assets/5759207/fcf558b2-3ff9-4539-ad9e-8455b52223a4
 
 `Mykyta (male)`:
 
-https://user-images.githubusercontent.com/5759207/224504363-0227e8bf-8c1c-49ad-8602-8cbf8feaa82b.mp4
+https://github.com/robinhad/ukrainian-tts/assets/5759207/033f5215-3f09-4021-ba19-1f55158445ca
+
 
 </details>
 

diff --git a/app.py b/app.py
@@ -43,6 +43,7 @@ class VoiceOption(Enum):
     Mykyta = "Микита (чоловічий) 👨"
     Lada = "Лада (жіночий) 👩"
     Dmytro = "Дмитро (чоловічий) 👨"
+    Oleksa = "Олекса (чоловічий) 👨"
 
 
 print(f"CUDA available? {is_available()}")
@@ -51,7 +52,7 @@ class VoiceOption(Enum):
 ukr_tts = TTS(device="cuda" if is_available() else "cpu")
 
 
-def tts(text: str, voice: str, speed: float):
+def tts(text: str, voice: str):
     print("============================")
     print("Original text:", text)
     print("Voice", voice)
@@ -62,6 +63,7 @@ def tts(text: str, voice: str, speed: float):
         VoiceOption.Mykyta.value: Voices.Mykyta.value,
         VoiceOption.Lada.value: Voices.Lada.value,
         VoiceOption.Dmytro.value: Voices.Dmytro.value,
+        VoiceOption.Oleksa.value: Voices.Oleksa.value,
     }
 
     speaker_name = voice_mapping[voice]
@@ -72,11 +74,11 @@ def tts(text: str, voice: str, speed: float):
 
     if getenv("HF_API_TOKEN") is not None:
         log_queue.put(
-            [text, speaker_name, Stress.Dictionary.value, speed, str(datetime.utcnow())]
+            [text, speaker_name, Stress.Dictionary.value, 1, str(datetime.utcnow())]
         )
 
     with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
-        _, text = ukr_tts.tts(text, speaker_name, Stress.Dictionary.value, fp, speed)
+        _, text = ukr_tts.tts(text, speaker_name, Stress.Dictionary.value, fp)
         return fp.name, text
 
 
@@ -97,9 +99,6 @@ def tts(text: str, voice: str, speed: float):
             choices=[option.value for option in VoiceOption],
             value=VoiceOption.Tetiana.value,
         ),
-        gr.components.Slider(
-            label="Швидкість", minimum=0.5, maximum=2, value=1, step=0.05
-        ),
     ],
     outputs=[
         gr.components.Audio(label="Output"),
@@ -112,32 +111,22 @@ def tts(text: str, voice: str, speed: float):
         [
             "Привіт, як тебе звати?",
             VoiceOption.Tetiana.value,
-            1,
         ],
         [
             "Введіть, будь ласка, св+оє реч+ення.",
             VoiceOption.Dmytro.value,
-            1,
-        ],
-        [
-            "Введіть, будь ласка, своє речення.",
-            VoiceOption.Dmytro.value,
-            1.3,
         ],
         [
             "Введіть, будь ласка, своє речення.",
-            VoiceOption.Mykyta.value,
-            1,
+            VoiceOption.Oleksa.value,
         ],
         [
             "Введіть, будь ласка, своє речення.",
             VoiceOption.Mykyta.value,
-            0.7,
         ],
         [
             "Договір підписано 4 квітня 1949 року.",
             VoiceOption.Lada.value,
-            0.9,
         ],
     ],
 )