Update app.py
Browse files
app.py
CHANGED
|
@@ -11,9 +11,6 @@ import random
|
|
| 11 |
from openai import OpenAI
|
| 12 |
import subprocess
|
| 13 |
from starlette.requests import ClientDisconnect
|
| 14 |
-
import hashlib
|
| 15 |
-
from functools import lru_cache
|
| 16 |
-
|
| 17 |
|
| 18 |
LLAMA_3B_API_ENDPOINT = os.environ.get("LLAMA_3B_API_ENDPOINT")
|
| 19 |
LLAMA_3B_API_KEY = os.environ.get("LLAMA_3B_API_KEY")
|
|
@@ -123,20 +120,16 @@ def models(text, model="Llama 3 8B Service", seed=42):
|
|
| 123 |
|
| 124 |
return output
|
| 125 |
|
| 126 |
-
|
|
|
|
|
|
|
| 127 |
|
| 128 |
-
@lru_cache(maxsize=100)
|
| 129 |
-
def cached_translate_speech(audio_file_hash, target_language):
|
| 130 |
-
# This function will be called if the result is not in the cache
|
| 131 |
language_code = LANGUAGE_CODES[target_language]
|
| 132 |
-
output_file =
|
| 133 |
-
|
| 134 |
-
if os.path.exists(output_file):
|
| 135 |
-
return output_file
|
| 136 |
|
| 137 |
command = [
|
| 138 |
"expressivity_predict",
|
| 139 |
-
|
| 140 |
"--tgt_lang", language_code,
|
| 141 |
"--model_name", "seamless_expressivity",
|
| 142 |
"--vocoder_name", "vocoder_pretssel",
|
|
@@ -144,39 +137,15 @@ def cached_translate_speech(audio_file_hash, target_language):
|
|
| 144 |
"--output_path", output_file
|
| 145 |
]
|
| 146 |
|
| 147 |
-
|
| 148 |
-
subprocess.run(command, check=True, capture_output=True, text=True)
|
| 149 |
-
if os.path.exists(output_file):
|
| 150 |
-
print(f"File created successfully: {output_file}")
|
| 151 |
-
return output_file
|
| 152 |
-
else:
|
| 153 |
-
print(f"File not found: {output_file}")
|
| 154 |
-
return None
|
| 155 |
-
except subprocess.CalledProcessError as e:
|
| 156 |
-
print(f"Translation process failed: {e.stderr}")
|
| 157 |
-
return None
|
| 158 |
|
| 159 |
-
|
| 160 |
-
|
|
|
|
|
|
|
|
|
|
| 161 |
return None
|
| 162 |
|
| 163 |
-
# Generate a hash of the audio file content
|
| 164 |
-
with open(audio_file, "rb") as f:
|
| 165 |
-
file_hash = hashlib.md5(f.read()).hexdigest()
|
| 166 |
-
|
| 167 |
-
# Check if we have a cached result
|
| 168 |
-
cache_key = (file_hash, target_language)
|
| 169 |
-
if cache_key in translation_cache:
|
| 170 |
-
return translation_cache[cache_key]
|
| 171 |
-
|
| 172 |
-
# If not in cache, call the cached function
|
| 173 |
-
result = cached_translate_speech(file_hash, target_language)
|
| 174 |
-
|
| 175 |
-
# Store the result in our custom cache
|
| 176 |
-
translation_cache[cache_key] = result
|
| 177 |
-
|
| 178 |
-
return result
|
| 179 |
-
|
| 180 |
async def respond(audio, model, seed, target_language):
|
| 181 |
try:
|
| 182 |
if audio is None:
|
|
@@ -212,7 +181,7 @@ def clear_history():
|
|
| 212 |
|
| 213 |
with gr.Blocks(css="style.css") as demo:
|
| 214 |
gr.Markdown("# <center><b>Optimus Prime: Your Personal AI Voice Assistant with Speech Translation</b></center>")
|
| 215 |
-
gr.Markdown("<center><b>For speech translation, start with the phrase 'Please translate' followed by the speech you want to translate</b></center>")
|
| 216 |
|
| 217 |
with gr.Row():
|
| 218 |
with gr.Column(scale=1):
|
|
|
|
| 11 |
from openai import OpenAI
|
| 12 |
import subprocess
|
| 13 |
from starlette.requests import ClientDisconnect
|
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
LLAMA_3B_API_ENDPOINT = os.environ.get("LLAMA_3B_API_ENDPOINT")
|
| 16 |
LLAMA_3B_API_KEY = os.environ.get("LLAMA_3B_API_KEY")
|
|
|
|
| 120 |
|
| 121 |
return output
|
| 122 |
|
| 123 |
+
def translate_speech(audio_file, target_language):
|
| 124 |
+
if audio_file is None:
|
| 125 |
+
return None
|
| 126 |
|
|
|
|
|
|
|
|
|
|
| 127 |
language_code = LANGUAGE_CODES[target_language]
|
| 128 |
+
output_file = "translated_audio.wav"
|
|
|
|
|
|
|
|
|
|
| 129 |
|
| 130 |
command = [
|
| 131 |
"expressivity_predict",
|
| 132 |
+
audio_file,
|
| 133 |
"--tgt_lang", language_code,
|
| 134 |
"--model_name", "seamless_expressivity",
|
| 135 |
"--vocoder_name", "vocoder_pretssel",
|
|
|
|
| 137 |
"--output_path", output_file
|
| 138 |
]
|
| 139 |
|
| 140 |
+
subprocess.run(command, check=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 141 |
|
| 142 |
+
if os.path.exists(output_file):
|
| 143 |
+
print(f"File created successfully: {output_file}")
|
| 144 |
+
return output_file
|
| 145 |
+
else:
|
| 146 |
+
print(f"File not found: {output_file}")
|
| 147 |
return None
|
| 148 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
async def respond(audio, model, seed, target_language):
|
| 150 |
try:
|
| 151 |
if audio is None:
|
|
|
|
| 181 |
|
| 182 |
with gr.Blocks(css="style.css") as demo:
|
| 183 |
gr.Markdown("# <center><b>Optimus Prime: Your Personal AI Voice Assistant with Speech Translation</b></center>")
|
| 184 |
+
gr.Markdown("## <center><b>For speech translation, start with the phrase 'Please translate' followed by the speech you want to translate</b></center>")
|
| 185 |
|
| 186 |
with gr.Row():
|
| 187 |
with gr.Column(scale=1):
|