zerogpu update

app.py

@@ -1,8 +1,11 @@
 import gradio as gr
 import requests
 from bs4 import BeautifulSoup
-import soundfile as sf
 from kokoro_tts import generate_audio
+import logging
+
+
+logging.basicConfig(level=logging.INFO)
 
 
 class Voices:

kokoro_tts.py

@@ -6,7 +6,7 @@ import logging
 
 def generate_audio(
     text,
-    lang_code="a",
+    pipeline=KPipeline(lang_code="a"),
     voice="af_heart",
     speed=1,
     save_segments=False,
@@ -25,7 +25,10 @@ def generate_audio(
     Returns:
         numpy.ndarray: Combined audio data at 24kHz sample rate
     """
-    pipeline = KPipeline(lang_code=lang_code)
+    try:
+        pipeline = pipeline.to("cuda")
+    except:
+        logging.warning("CUDA not available, using CPU")
 
     generator = pipeline(text, voice=voice, speed=speed, split_pattern=r"\.")
 

pyproject.toml

@@ -10,4 +10,5 @@ dependencies = [
     "kokoro>=0.3.4",
     "pip>=25.0.1",
     "soundfile>=0.13.1",
+    "torch==2.4.0",
 ]

requirements.txt

@@ -21,9 +21,7 @@ curated-transformers==0.1.1
 cymem==2.0.11
 dlinfo==2.0.0
 docopt==0.6.2
-en_core_web_sm @ https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.8.0/en_core_web_sm-3.8.0-py3-none-any.whl#sha256=1932429db727d4bff3deed6b34cfc05df17794f4a52eeb26cf8928f7c1a0fb85
 espeakng-loader==0.2.4
-exceptiongroup==1.2.2
 fastapi==0.115.8
 ffmpy==0.5.0
 filelock==3.17.0
@@ -87,6 +85,7 @@ safetensors==0.5.2
 scipy==1.15.1
 segments==2.2.1
 semantic-version==2.10.0
+setuptools==75.8.0
 shellingham==1.5.4
 six==1.17.0
 smart-open==7.1.0
@@ -99,12 +98,12 @@ spacy-legacy==3.0.12
 spacy-loggers==1.0.5
 srsly==2.5.1
 starlette==0.45.3
-sympy==1.13.1
+sympy==1.13.3
 tabulate==0.9.0
 thinc==8.3.4
 tokenizers==0.21.0
 tomlkit==0.13.2
-torch==2.6.0
+torch==2.4.0
 tqdm==4.67.1
 transformers==4.48.3
 typer==0.15.1