Spaces:
Sleeping
Sleeping
Commit
·
ff30a18
1
Parent(s):
50e0de0
[ADD] app.py files
Browse files- .idea/.gitignore +5 -0
- .idea/inspectionProfiles/Project_Default.xml +32 -0
- .idea/inspectionProfiles/profiles_settings.xml +6 -0
- .idea/misc.xml +7 -0
- .idea/modules.xml +8 -0
- .idea/stt-fongbe.iml +8 -0
- .idea/vcs.xml +7 -0
- app.py +176 -0
.idea/.gitignore
ADDED
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Default ignored files
|
2 |
+
/shelf/
|
3 |
+
/workspace.xml
|
4 |
+
# Editor-based HTTP Client requests
|
5 |
+
/httpRequests/
|
.idea/inspectionProfiles/Project_Default.xml
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<component name="InspectionProjectProfileManager">
|
2 |
+
<profile version="1.0">
|
3 |
+
<option name="myName" value="Project Default" />
|
4 |
+
<inspection_tool class="PyPackageRequirementsInspection" enabled="true" level="WARNING" enabled_by_default="true">
|
5 |
+
<option name="ignoredPackages">
|
6 |
+
<list>
|
7 |
+
<option value="aiohappyeyeballs" />
|
8 |
+
<option value="aiohttp" />
|
9 |
+
<option value="aiosignal" />
|
10 |
+
<option value="annotated-types" />
|
11 |
+
<option value="anyio" />
|
12 |
+
<option value="attrs" />
|
13 |
+
<option value="click" />
|
14 |
+
<option value="fastapi" />
|
15 |
+
<option value="frozenlist" />
|
16 |
+
<option value="h11" />
|
17 |
+
<option value="idna" />
|
18 |
+
<option value="multidict" />
|
19 |
+
<option value="propcache" />
|
20 |
+
<option value="pydantic" />
|
21 |
+
<option value="pydantic_core" />
|
22 |
+
<option value="sniffio" />
|
23 |
+
<option value="starlette" />
|
24 |
+
<option value="typing-inspection" />
|
25 |
+
<option value="typing_extensions" />
|
26 |
+
<option value="uvicorn" />
|
27 |
+
<option value="yarl" />
|
28 |
+
</list>
|
29 |
+
</option>
|
30 |
+
</inspection_tool>
|
31 |
+
</profile>
|
32 |
+
</component>
|
.idea/inspectionProfiles/profiles_settings.xml
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<component name="InspectionProjectProfileManager">
|
2 |
+
<settings>
|
3 |
+
<option name="USE_PROJECT_PROFILE" value="false" />
|
4 |
+
<version value="1.0" />
|
5 |
+
</settings>
|
6 |
+
</component>
|
.idea/misc.xml
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<project version="4">
|
3 |
+
<component name="Black">
|
4 |
+
<option name="sdkName" value="Python 3.11 (stt-fongbe)" />
|
5 |
+
</component>
|
6 |
+
<component name="ProjectRootManager" version="2" project-jdk-name="Python 3.11 (stt-fongbe)" project-jdk-type="Python SDK" />
|
7 |
+
</project>
|
.idea/modules.xml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<project version="4">
|
3 |
+
<component name="ProjectModuleManager">
|
4 |
+
<modules>
|
5 |
+
<module fileurl="file://$PROJECT_DIR$/.idea/stt-fongbe.iml" filepath="$PROJECT_DIR$/.idea/stt-fongbe.iml" />
|
6 |
+
</modules>
|
7 |
+
</component>
|
8 |
+
</project>
|
.idea/stt-fongbe.iml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<module type="PYTHON_MODULE" version="4">
|
3 |
+
<component name="NewModuleRootManager">
|
4 |
+
<content url="file://$MODULE_DIR$" />
|
5 |
+
<orderEntry type="jdk" jdkName="Python 3.11 (stt-fongbe)" jdkType="Python SDK" />
|
6 |
+
<orderEntry type="sourceFolder" forTests="false" />
|
7 |
+
</component>
|
8 |
+
</module>
|
.idea/vcs.xml
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<project version="4">
|
3 |
+
<component name="VcsDirectoryMappings">
|
4 |
+
<mapping directory="" vcs="Git" />
|
5 |
+
<mapping directory="$PROJECT_DIR$" vcs="Git" />
|
6 |
+
</component>
|
7 |
+
</project>
|
app.py
ADDED
@@ -0,0 +1,176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
import torchaudio
|
4 |
+
import librosa
|
5 |
+
import os
|
6 |
+
from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
|
7 |
+
from huggingface_hub import login
|
8 |
+
import logging
|
9 |
+
|
10 |
+
# Configuration
|
11 |
+
MODEL_NAME = "Ronaldodev/speech-to-text-fongbe"
|
12 |
+
HF_TOKEN = os.environ.get("HF_TOKEN")
|
13 |
+
|
14 |
+
# Variables globales
|
15 |
+
model = None
|
16 |
+
processor = None
|
17 |
+
|
18 |
+
# Configuration logging
|
19 |
+
logging.basicConfig(level=logging.INFO)
|
20 |
+
logger = logging.getLogger(__name__)
|
21 |
+
|
22 |
+
|
23 |
+
def load_model():
|
24 |
+
"""Charger le modèle privé au démarrage"""
|
25 |
+
global model, processor
|
26 |
+
|
27 |
+
try:
|
28 |
+
logger.info("🔄 Chargement du modèle privé...")
|
29 |
+
|
30 |
+
if not HF_TOKEN:
|
31 |
+
raise ValueError("HF_TOKEN non configuré dans les secrets")
|
32 |
+
|
33 |
+
# Login avec token privé
|
34 |
+
login(token=HF_TOKEN)
|
35 |
+
logger.info("✅ Authentification HF réussie")
|
36 |
+
|
37 |
+
# Charger le modèle et processeur
|
38 |
+
model = AutoModelForSpeechSeq2Seq.from_pretrained(MODEL_NAME)
|
39 |
+
processor = AutoProcessor.from_pretrained(MODEL_NAME)
|
40 |
+
|
41 |
+
logger.info("✅ Modèle chargé avec succès!")
|
42 |
+
return True
|
43 |
+
|
44 |
+
except Exception as e:
|
45 |
+
logger.error(f"❌ Erreur chargement: {e}")
|
46 |
+
return False
|
47 |
+
|
48 |
+
|
49 |
+
def transcribe(audio):
|
50 |
+
"""Fonction principale de transcription"""
|
51 |
+
|
52 |
+
# Vérifier si le modèle est chargé
|
53 |
+
if model is None or processor is None:
|
54 |
+
return "❌ Erreur: Modèle non chargé. Vérifiez les logs."
|
55 |
+
|
56 |
+
# Vérifier si un audio est fourni
|
57 |
+
if audio is None:
|
58 |
+
return "❌ Aucun fichier audio fourni"
|
59 |
+
|
60 |
+
try:
|
61 |
+
logger.info(f"🎵 Traitement audio: {audio}")
|
62 |
+
|
63 |
+
# Charger l'audio avec fallback
|
64 |
+
try:
|
65 |
+
waveform, sample_rate = torchaudio.load(audio)
|
66 |
+
logger.info(f"✅ Audio chargé avec torchaudio: {sample_rate}Hz")
|
67 |
+
except Exception as e:
|
68 |
+
logger.warning(f"⚠️ Torchaudio échoué, essai librosa: {e}")
|
69 |
+
waveform, sample_rate = librosa.load(audio, sr=None)
|
70 |
+
waveform = torch.tensor(waveform).unsqueeze(0)
|
71 |
+
logger.info(f"✅ Audio chargé avec librosa: {sample_rate}Hz")
|
72 |
+
|
73 |
+
# Conversion mono si nécessaire
|
74 |
+
if waveform.shape[0] > 1:
|
75 |
+
waveform = waveform.mean(dim=0, keepdim=True)
|
76 |
+
logger.info("🔄 Conversion stéréo → mono")
|
77 |
+
|
78 |
+
# Resampling à 16kHz si nécessaire
|
79 |
+
if sample_rate != 16000:
|
80 |
+
logger.info(f"🔄 Resampling {sample_rate}Hz → 16000Hz")
|
81 |
+
resampler = torchaudio.transforms.Resample(sample_rate, 16000)
|
82 |
+
waveform = resampler(waveform)
|
83 |
+
|
84 |
+
# Préparation des inputs
|
85 |
+
inputs = processor(
|
86 |
+
waveform.squeeze(),
|
87 |
+
sampling_rate=16000,
|
88 |
+
return_tensors="pt"
|
89 |
+
)
|
90 |
+
|
91 |
+
# Génération de la transcription
|
92 |
+
logger.info("🔄 Génération de la transcription...")
|
93 |
+
with torch.no_grad():
|
94 |
+
result = model.generate(
|
95 |
+
**inputs,
|
96 |
+
max_length=500,
|
97 |
+
do_sample=False,
|
98 |
+
num_beams=1
|
99 |
+
)
|
100 |
+
|
101 |
+
# Décodage
|
102 |
+
transcription = processor.batch_decode(result, skip_special_tokens=True)[0]
|
103 |
+
|
104 |
+
logger.info(f"✅ Transcription réussie: '{transcription}'")
|
105 |
+
return transcription.strip()
|
106 |
+
|
107 |
+
except Exception as e:
|
108 |
+
error_msg = f"❌ Erreur de transcription: {str(e)}"
|
109 |
+
logger.error(error_msg)
|
110 |
+
return error_msg
|
111 |
+
|
112 |
+
|
113 |
+
# Charger le modèle au démarrage
|
114 |
+
print("🚀 DÉMARRAGE API STT FONGBÉ - RONALDODEV")
|
115 |
+
print("=" * 50)
|
116 |
+
|
117 |
+
if load_model():
|
118 |
+
print("✅ Modèle chargé - Interface prête!")
|
119 |
+
model_status = "✅ Modèle chargé et prêt"
|
120 |
+
else:
|
121 |
+
print("❌ Erreur de chargement du modèle")
|
122 |
+
model_status = "❌ Erreur de chargement"
|
123 |
+
|
124 |
+
# Interface Gradio simple
|
125 |
+
demo = gr.Interface(
|
126 |
+
fn=transcribe,
|
127 |
+
inputs=gr.Audio(
|
128 |
+
sources=["upload", "microphone"],
|
129 |
+
type="filepath",
|
130 |
+
label="🎤 Uploadez un fichier ou enregistrez directement"
|
131 |
+
),
|
132 |
+
outputs=gr.Textbox(
|
133 |
+
label="📝 Transcription en Fongbé",
|
134 |
+
placeholder="La transcription apparaîtra ici...",
|
135 |
+
lines=3
|
136 |
+
),
|
137 |
+
title="🎤 API STT Fongbé - Ronaldodev",
|
138 |
+
description=f"""
|
139 |
+
**Reconnaissance vocale pour la langue Fongbé**
|
140 |
+
|
141 |
+
Uploadez un fichier audio (WAV, MP3, M4A) ou enregistrez directement avec votre microphone.
|
142 |
+
|
143 |
+
**Statut:** {model_status}
|
144 |
+
|
145 |
+
**Modèle:** `{MODEL_NAME}`
|
146 |
+
""",
|
147 |
+
article="""
|
148 |
+
## 🔌 API pour développeurs
|
149 |
+
|
150 |
+
Cette interface expose automatiquement une API REST :
|
151 |
+
|
152 |
+
**Endpoint:** `POST /api/predict`
|
153 |
+
|
154 |
+
**Exemple d'utilisation:**
|
155 |
+
```python
|
156 |
+
import requests
|
157 |
+
|
158 |
+
response = requests.post(
|
159 |
+
"https://ronaldodev-stt-fongbe.hf.space/api/predict",
|
160 |
+
json={"data": [audio_file_path]}
|
161 |
+
)
|
162 |
+
|
163 |
+
transcription = response.json()["data"][0]
|
164 |
+
```
|
165 |
+
|
166 |
+
**Pour Flutter:** Utilisez MultipartRequest avec l'endpoint ci-dessus.
|
167 |
+
""",
|
168 |
+
examples=[
|
169 |
+
# Vous pouvez ajouter des fichiers d'exemple si vous en avez
|
170 |
+
],
|
171 |
+
theme=gr.themes.Soft(),
|
172 |
+
allow_flagging="never"
|
173 |
+
)
|
174 |
+
|
175 |
+
# Lancement de l'interface
|
176 |
+
demo.launch()
|