Update app.py
Browse files
app.py
CHANGED
|
@@ -8,12 +8,12 @@ from openai import OpenAI
|
|
| 8 |
st.set_page_config(page_title="Urdu TTS - OpenAI", page_icon="🔊", layout="centered")
|
| 9 |
|
| 10 |
st.title("🔊 Urdu Text → Speech (OpenAI)")
|
| 11 |
-
st.caption("Type Urdu text and generate natural Urdu speech with OpenAI TTS. If you have a custom
|
| 12 |
|
| 13 |
-
# ---- API key ----
|
| 14 |
-
API_KEY = os.getenv("
|
| 15 |
if not API_KEY:
|
| 16 |
-
st.error("Missing
|
| 17 |
st.stop()
|
| 18 |
|
| 19 |
client = OpenAI(api_key=API_KEY)
|
|
@@ -21,45 +21,34 @@ client = OpenAI(api_key=API_KEY)
|
|
| 21 |
# ---- Sidebar options ----
|
| 22 |
with st.sidebar:
|
| 23 |
st.header("Options")
|
| 24 |
-
|
| 25 |
-
voice_presets = ["alloy", "verse", "aria", "ballad", "cove", "luna", "sage"]
|
| 26 |
use_custom = st.checkbox("Use custom OpenAI voice ID", False)
|
| 27 |
custom_voice = ""
|
| 28 |
if use_custom:
|
| 29 |
-
custom_voice = st.text_input("Custom voice_id", value="", help="
|
| 30 |
else:
|
| 31 |
-
preset = st.selectbox("Built-in voice", options=
|
| 32 |
out_name = st.text_input("Output filename (no extension)", "urdu_tts")
|
| 33 |
-
fmt = st.selectbox("Audio format", options=["
|
| 34 |
|
| 35 |
-
|
| 36 |
-
|
|
|
|
| 37 |
|
| 38 |
col1, col2 = st.columns(2)
|
| 39 |
with col1:
|
| 40 |
-
|
| 41 |
with col2:
|
| 42 |
-
|
| 43 |
|
| 44 |
-
if
|
| 45 |
st.session_state.pop("audio_bytes", None)
|
| 46 |
st.experimental_rerun()
|
| 47 |
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
Uses OpenAI TTS (gpt-4o-mini-tts) to synthesize speech.
|
| 51 |
-
We write to a temp file using streaming response, then return bytes.
|
| 52 |
-
"""
|
| 53 |
model = "gpt-4o-mini-tts"
|
| 54 |
-
|
| 55 |
-
# mp3_44100_128 is recommended for quality-size balance
|
| 56 |
-
if output_format == "wav":
|
| 57 |
-
audio_format = "wav"
|
| 58 |
-
ext = "wav"
|
| 59 |
-
else:
|
| 60 |
-
audio_format = "mp3"
|
| 61 |
-
ext = "mp3"
|
| 62 |
-
|
| 63 |
tmp_path = Path(f"/tmp/tts_{datetime.now().strftime('%H%M%S')}.{ext}")
|
| 64 |
with client.audio.speech.with_streaming_response.create(
|
| 65 |
model=model,
|
|
@@ -68,7 +57,6 @@ def tts_openai(urdu_text: str, voice: str, output_format: str) -> bytes:
|
|
| 68 |
format=audio_format
|
| 69 |
) as resp:
|
| 70 |
resp.stream_to_file(tmp_path)
|
| 71 |
-
|
| 72 |
data = tmp_path.read_bytes()
|
| 73 |
try:
|
| 74 |
tmp_path.unlink(missing_ok=True)
|
|
@@ -76,7 +64,8 @@ def tts_openai(urdu_text: str, voice: str, output_format: str) -> bytes:
|
|
| 76 |
pass
|
| 77 |
return data, ext
|
| 78 |
|
| 79 |
-
|
|
|
|
| 80 |
if not text.strip():
|
| 81 |
st.warning("براہ کرم اردو متن درج کریں")
|
| 82 |
else:
|
|
@@ -85,7 +74,7 @@ if make_audio:
|
|
| 85 |
if not voice_to_use:
|
| 86 |
st.warning("Custom voice ID is empty. Either uncheck custom voice or provide a valid voice_id.")
|
| 87 |
else:
|
| 88 |
-
st.info("Generating speech with OpenAI TTS…")
|
| 89 |
audio_bytes, ext = tts_openai(text.strip(), voice_to_use, fmt)
|
| 90 |
st.session_state["audio_bytes"] = audio_bytes
|
| 91 |
st.session_state["ext"] = ext
|
|
@@ -93,7 +82,7 @@ if make_audio:
|
|
| 93 |
except Exception as e:
|
| 94 |
st.error(f"کچھ مسئلہ آیا: {e}")
|
| 95 |
|
| 96 |
-
# ---- Preview
|
| 97 |
if "audio_bytes" in st.session_state:
|
| 98 |
ext = st.session_state.get("ext", "mp3")
|
| 99 |
st.markdown("### ▶️ Preview")
|
|
@@ -109,6 +98,6 @@ if "audio_bytes" in st.session_state:
|
|
| 109 |
|
| 110 |
st.markdown("---")
|
| 111 |
st.caption(
|
| 112 |
-
"Notes: Built-in voices are not your personal voice. For
|
| 113 |
-
"Urdu is supported by gpt-4o-mini-tts.
|
| 114 |
)
|
|
|
|
| 8 |
st.set_page_config(page_title="Urdu TTS - OpenAI", page_icon="🔊", layout="centered")
|
| 9 |
|
| 10 |
st.title("🔊 Urdu Text → Speech (OpenAI)")
|
| 11 |
+
st.caption("Type Urdu text and generate natural Urdu speech with OpenAI TTS. If you have a custom voice ID, you can use it.")
|
| 12 |
|
| 13 |
+
# ---- API key from Hugging Face Secret ----
|
| 14 |
+
API_KEY = os.getenv("Key_1") or st.secrets.get("Key_1")
|
| 15 |
if not API_KEY:
|
| 16 |
+
st.error("Missing Key_1. Go to Settings → Secrets in your Space and add it.")
|
| 17 |
st.stop()
|
| 18 |
|
| 19 |
client = OpenAI(api_key=API_KEY)
|
|
|
|
| 21 |
# ---- Sidebar options ----
|
| 22 |
with st.sidebar:
|
| 23 |
st.header("Options")
|
| 24 |
+
voices = ["alloy", "verse", "aria", "ballad", "cove", "luna", "sage"]
|
|
|
|
| 25 |
use_custom = st.checkbox("Use custom OpenAI voice ID", False)
|
| 26 |
custom_voice = ""
|
| 27 |
if use_custom:
|
| 28 |
+
custom_voice = st.text_input("Custom voice_id", value="", help="Only works if your account has custom voices enabled")
|
| 29 |
else:
|
| 30 |
+
preset = st.selectbox("Built-in voice", options=voices, index=0)
|
| 31 |
out_name = st.text_input("Output filename (no extension)", "urdu_tts")
|
| 32 |
+
fmt = st.selectbox("Audio format", options=["mp3", "wav"], index=0)
|
| 33 |
|
| 34 |
+
# ---- Main input ----
|
| 35 |
+
default_text = "یہ ایک سادہ مثال ہے۔ یہاں اپنا متن لکھیں اور آڈیو حاصل کریں۔"
|
| 36 |
+
text = st.text_area("Urdu text", value=default_text, height=200, placeholder="یہاں اردو میں ٹیکسٹ لکھیں یا پیسٹ کریں…")
|
| 37 |
|
| 38 |
col1, col2 = st.columns(2)
|
| 39 |
with col1:
|
| 40 |
+
run_btn = st.button("🎙️ Generate", use_container_width=True)
|
| 41 |
with col2:
|
| 42 |
+
clear_btn = st.button("🧹 Clear", use_container_width=True)
|
| 43 |
|
| 44 |
+
if clear_btn:
|
| 45 |
st.session_state.pop("audio_bytes", None)
|
| 46 |
st.experimental_rerun()
|
| 47 |
|
| 48 |
+
# ---- Helper function ----
|
| 49 |
+
def tts_openai(urdu_text: str, voice: str, audio_format: str) -> bytes:
|
|
|
|
|
|
|
|
|
|
| 50 |
model = "gpt-4o-mini-tts"
|
| 51 |
+
ext = "mp3" if audio_format == "mp3" else "wav"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
tmp_path = Path(f"/tmp/tts_{datetime.now().strftime('%H%M%S')}.{ext}")
|
| 53 |
with client.audio.speech.with_streaming_response.create(
|
| 54 |
model=model,
|
|
|
|
| 57 |
format=audio_format
|
| 58 |
) as resp:
|
| 59 |
resp.stream_to_file(tmp_path)
|
|
|
|
| 60 |
data = tmp_path.read_bytes()
|
| 61 |
try:
|
| 62 |
tmp_path.unlink(missing_ok=True)
|
|
|
|
| 64 |
pass
|
| 65 |
return data, ext
|
| 66 |
|
| 67 |
+
# ---- Generate speech ----
|
| 68 |
+
if run_btn:
|
| 69 |
if not text.strip():
|
| 70 |
st.warning("براہ کرم اردو متن درج کریں")
|
| 71 |
else:
|
|
|
|
| 74 |
if not voice_to_use:
|
| 75 |
st.warning("Custom voice ID is empty. Either uncheck custom voice or provide a valid voice_id.")
|
| 76 |
else:
|
| 77 |
+
st.info("Generating Urdu speech with OpenAI TTS…")
|
| 78 |
audio_bytes, ext = tts_openai(text.strip(), voice_to_use, fmt)
|
| 79 |
st.session_state["audio_bytes"] = audio_bytes
|
| 80 |
st.session_state["ext"] = ext
|
|
|
|
| 82 |
except Exception as e:
|
| 83 |
st.error(f"کچھ مسئلہ آیا: {e}")
|
| 84 |
|
| 85 |
+
# ---- Preview & download ----
|
| 86 |
if "audio_bytes" in st.session_state:
|
| 87 |
ext = st.session_state.get("ext", "mp3")
|
| 88 |
st.markdown("### ▶️ Preview")
|
|
|
|
| 98 |
|
| 99 |
st.markdown("---")
|
| 100 |
st.caption(
|
| 101 |
+
"Notes: Built-in voices are not your personal voice. For your own cloned voice, you need access to custom OpenAI Voice IDs. "
|
| 102 |
+
"Urdu is supported by gpt-4o-mini-tts."
|
| 103 |
)
|