rework organization of project

app.py

@@ -12,12 +12,11 @@ import torch
 from huggingface_hub import snapshot_download
 import numpy as np
 
-from omnivoice import (
-    OmniVoice,
-    OmniVoiceGenerationConfig,
-)
-#from f5_tts.api import F5TTS
-import tempfile
+from config.settings import *
+from data.contexts import CONTEXTS
+from data.dataset_manager import *
+from services import audio_service
+
 
 timestamp = datetime.now(UTC).isoformat()
 
@@ -29,219 +28,7 @@ SENTENCES = [
     "Les systèmes vocaux progressent rapidement"
 ]
 
-CONTEXTS = {
-"Train station announcement":
-"Le train 7891 à destination de Paris Nord, prévu à 14h35, partira voie 8.",
- 
-"Weather forecast":
-"Demain, les températures atteindront 27 degrés avec un ciel dégagé.",
- 
-"Voice assistant":
-"Je peux vous aider à trouver le restaurant le plus proche.",
- 
-"Audiobook":
-"Le vieux château se dressait au sommet de la colline depuis plusieurs siècles.",
- 
-"Customer service":
-"Votre demande a bien été prise en compte et sera traitée sous quarante-huit heures."
-}
-
-# ============================
-# CONFIGURATION
-# ============================
-
-ADMIN_PASSWORD_HASH = "eb32da077dfaf326cd6f73e0716b628da6427aa318a2d0b9fafa9ef315b5e885"
-
-# ============================
-# HF DATASET CONFIGURATION
-# ============================
-
-HF_DATASET_NAME = "PaulineDV/TTS_annotations_data"  # replace with your HF dataset
-HF_TOKEN = os.environ.get("MyJulySecretToken")  # store your token as a secret in Spaces
-login(HF_TOKEN)
-
-# ============================
-# MODEL AND GENERATING AUDIO
-# ============================
-
-REFERENCE_AUDIO = "references/basic_ref_en.wav"
-
-REFERENCE_TEXT = (
-    "Some call me nature, others call me mother nature."
-)
-
-MODEL_OPTIONS = [
-    "OmniVoice",
-    "F5-TTS",
-    "Qwen3-TTS"
-]
-
-print("Loading OmniVoice...")
-
-tts_model = OmniVoice.from_pretrained(
-    "k2-fsa/OmniVoice",
-    device_map="cpu",
-    dtype=torch.float32
-)
-
-SAMPLING_RATE = tts_model.sampling_rate
-
-print("OmniVoice loaded")
-
-print("Loading F5-TTS...")
-
-#f5tts = F5TTS()
-
-print("F5-TTS loaded")
-
-def generate_audio(model_name, context):
-
-    sentence = CONTEXTS[context]
-
-    if model_name == "OmniVoice":
-        return generate_omnivoice(sentence)
-    
-    elif model_name == "F5-TTS":
-        return generate_f5(sentence)
-    
-    #elif model_name == "Qwen3-TTS":
-    #    return generate_qwen(sentence)
-    
-def generate_omnivoice(sentence):
-
-    config = OmniVoiceGenerationConfig(
-        num_step=32,
-        guidance_scale=2.0
-    )
-
-    audio = tts_model.generate(
-        text=sentence,
-        language="French",
-        generation_config=config
-    )
-
-    #convertir l'array en fichier audio
-    waveform = (audio[0] * 32767).astype(np.int16)
-
-    return (SAMPLING_RATE, waveform), {"played": 0}
-
-def generate_f5(sentence):
-    
-    output_wav_path = tempfile.mktemp(
-        suffix = ".wav"
-    )
-
-    wav, sr, _ = f5tts.infer(
-        ref_file = REFERENCE_AUDIO,
-        ref_text = REFERENCE_TEXT,
-        gen_text = sentence,
-        file_wave = output_wav_path,
-        remove_silence = False,
-    )
-
-    return output_wav_path, {"played": 0}
-
-# ============================
-# USER FUNCTIONS
-# ============================
-api = HfApi(token = HF_TOKEN)
-
-def load_hf_dataset():
-    """Load existing HF Dataset or create empty one if not exists."""
-    try:
-        ds = load_dataset(HF_DATASET_NAME, split="train")
-    except:
-        # Dataset does not exist yet
-        df = pd.DataFrame(columns=["user_id", "gender", "audio_file", "score"])
-        ds = Dataset.from_pandas(df)
-        ds.push_to_hub(HF_DATASET_NAME, private=True)
-    return ds
-
-def submit_annotation_hf(
-        user_id, 
-        age_group, 
-        gender, 
-        native_language, 
-        tts_experience, 
-        device_type, 
-        selected_model,
-        context, 
-        score,
-        context_score
-):
-    annotation = {
-        "user_id": user_id,
-        "age_group": age_group,
-        "gender": gender,
-        "native_language": native_language,
-        "tts_experience": tts_experience,
-        "device_type": device_type,
-        
-        "model_name": selected_model,
-        "context": context,
-
-        "score": score,
-        "context_score": context_score,
-        "timestamp": datetime.now(UTC).isoformat()
-    }
-
-    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
-    short_id = str(uuid.uuid4()) [:8]
-
-    file_name = (
-        f"{selected_model}_{user_id}_{timestamp}_{short_id}"
-    )
-
-    local_file = f"{file_name}.json"
-
-    with open(local_file, "w", encoding="utf-8") as f:
-        json.dump(annotation, f, indent=2)
-
-    api.upload_file(
-        path_or_fileobj=local_file,
-        path_in_repo=f"annotations/{local_file}",
-        repo_id=HF_DATASET_NAME,
-        repo_type="dataset"
-    )
-    
-    os.remove(local_file)
-
-
-def submit_annotation(
-        user_id, 
-        age_group, 
-        gender, 
-        native_language, 
-        tts_experience, 
-        device_type, 
-        selected_model,
-        context, 
-        score,
-        context_score
-):
-   try:
-
-    submit_annotation_hf(
-        user_id, 
-        age_group, 
-        gender, 
-        native_language, 
-        tts_experience, 
-        device_type, 
-        selected_model,
-        context, 
-        score,
-        context_score
-    )
-
-    return f"Annotation saved for {context}."
-
-   except Exception as e:
-       print("Submit error")
-       print(type(e).__name__)
-       print(e)
 
-       return f"ERROR: {e}"
 
 def update_sentence(context):
 

config/settings.py

@@ -0,0 +1,15 @@
+HF_DATASET_NAME = "PaulineDV/TTS_annotations_data" 
+
+REFERENCE_AUDIO = "references/basic_ref_en.wav"
+
+REFERENCE_TEXT = (
+    "Some call me nature, others call me mother nature."
+)
+
+MODEL_OPTIONS = [
+    "OmniVoice",
+    "F5-TTS",
+    "Qwen3-TTS"
+]
+
+ADMIN_PASSWORD_HASH = "eb32da077dfaf326cd6f73e0716b628da6427aa318a2d0b9fafa9ef315b5e885"

data/contexts.py

@@ -0,0 +1,16 @@
+CONTEXTS = {
+"Train station announcement":
+"Le train 7891 à destination de Paris Nord, prévu à 14h35, partira voie 8.",
+ 
+"Weather forecast":
+"Demain, les températures atteindront 27 degrés avec un ciel dégagé.",
+ 
+"Voice assistant":
+"Je peux vous aider à trouver le restaurant le plus proche.",
+ 
+"Audiobook":
+"Le vieux château se dressait au sommet de la colline depuis plusieurs siècles.",
+ 
+"Customer service":
+"Votre demande a bien été prise en compte et sera traitée sous quarante-huit heures."
+}

data/dataset_manager.py

@@ -0,0 +1,34 @@
+from huggingface_hub import HfApi
+import gradio as gr
+import pandas as pd
+import os
+import hashlib
+from datasets import Dataset, concatenate_datasets, load_dataset
+from huggingface_hub import login
+import json
+import uuid
+from datetime import datetime, UTC
+import torch
+from huggingface_hub import snapshot_download
+import numpy as np
+
+from config.settings import *
+from data.contexts import CONTEXTS
+
+HF_TOKEN = os.environ.get("MyJulySecretToken")  # store your token as a secret in Spaces
+login(HF_TOKEN)
+
+
+
+api = HfApi(token = HF_TOKEN)
+
+def load_hf_dataset():
+    """Load existing HF Dataset or create empty one if not exists."""
+    try:
+        ds = load_dataset(HF_DATASET_NAME, split="train")
+    except:
+        # Dataset does not exist yet
+        df = pd.DataFrame(columns=["user_id", "gender", "audio_file", "score"])
+        ds = Dataset.from_pandas(df)
+        ds.push_to_hub(HF_DATASET_NAME, private=True)
+    return ds

models/f5tts_model.py

@@ -0,0 +1,29 @@
+from f5_tts.api import F5TTS
+import tempfile
+
+from config.settings import (
+    REFERENCE_AUDIO,
+    REFERENCE_TEXT
+)
+
+print("Loading F5-TTS...")
+
+model = F5TTS()
+
+print("F5-TTS loaded")
+
+def generate(sentence):
+    
+    output_wav = tempfile.mkstemp(
+        suffix = ".wav"
+    )
+
+    model.infer(
+        ref_file = REFERENCE_AUDIO,
+        ref_text = REFERENCE_TEXT,
+        gen_text = sentence,
+        file_wave = output_wav,
+        remove_silence = False,
+    )
+
+    return output_wav

models/manager.py

@@ -0,0 +1,15 @@
+from models.omnivoice_model import generate as generate_omnivoice
+from models.f5tts_model import generate as generate_f5
+
+def generate_audio(model_name, sentence):
+
+    if model_name == "OmniVoice":
+        return generate_omnivoice(sentence)
+    
+    elif model_name == "F5-TTS":
+        return generate_f5(sentence)
+    
+    raise ValueError(
+        f"Unknown model: {model_name}"
+    )
+    

models/omnivoice_model.py

@@ -0,0 +1,38 @@
+import numpy as np
+import torch
+
+from omnivoice import (
+    OmniVoice,
+    OmniVoiceGenerationConfig,
+)
+
+print("Loading OmniVoice...")
+
+tts_model = OmniVoice.from_pretrained(
+    "k2-fsa/OmniVoice",
+    device_map="cpu",
+    dtype=torch.float32
+)
+
+SAMPLING_RATE = tts_model.sampling_rate
+
+print("OmniVoice loaded")
+
+def generate(sentence):
+
+    config = OmniVoiceGenerationConfig(
+        num_step=32,
+        guidance_scale=2.0
+    )
+
+    audio = tts_model.generate(
+        text=sentence,
+        language="French",
+        generation_config=config
+    )
+
+    #convertir l'array en fichier audio
+    waveform = (audio[0] * 32767).astype(np.int16)
+
+    return (SAMPLING_RATE, waveform)
+

services/annotation_service.py

@@ -0,0 +1,127 @@
+from huggingface_hub import HfApi
+import gradio as gr
+import pandas as pd
+import os
+from datetime import datetime, UTC
+from datasets import Dataset, concatenate_datasets, load_dataset
+from huggingface_hub import login
+import json
+import uuid
+from datetime import datetime, UTC
+import torch
+from huggingface_hub import snapshot_download
+
+from config.settings import *
+
+HF_TOKEN = os.environ.get("MyJulySecretToken")  # store your token as a secret in Spaces
+login(HF_TOKEN)
+
+HF_DATASET_NAME = "PaulineDV/TTS_annotations_data" 
+
+api = HfApi(token = HF_TOKEN)
+
+def load_hf_dataset():
+    """Load existing HF Dataset or create empty one if not exists."""
+    try:
+        ds = load_dataset(HF_DATASET_NAME, split="train")
+    except:
+        # Dataset does not exist yet
+        df = pd.DataFrame(columns=["user_id", "gender", "audio_file", "score"])
+        ds = Dataset.from_pandas(df)
+        ds.push_to_hub(HF_DATASET_NAME, private=True)
+    return ds
+
+
+
+timestamp = datetime.now(UTC).isoformat()
+
+
+def submit_annotation_hf(
+        user_id, 
+        age_group, 
+        gender, 
+        native_language, 
+        tts_experience, 
+        device_type, 
+        selected_model,
+        context, 
+        score,
+        context_score
+):
+    annotation = {
+        "user_id": user_id,
+        "age_group": age_group,
+        "gender": gender,
+        "native_language": native_language,
+        "tts_experience": tts_experience,
+        "device_type": device_type,
+        
+        "model_name": selected_model,
+        "context": context,
+
+        "score": score,
+        "context_score": context_score,
+        "timestamp": datetime.now(UTC).isoformat()
+    }
+
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    short_id = str(uuid.uuid4()) [:8]
+
+    file_name = (
+        f"{selected_model}_{user_id}_{timestamp}_{short_id}"
+    )
+
+    local_file = f"{file_name}.json"
+
+    with open(local_file, "w", encoding="utf-8") as f:
+        json.dump(annotation, f, indent=2)
+
+    api.upload_file(
+        path_or_fileobj=local_file,
+        path_in_repo=f"annotations/{local_file}",
+        repo_id=HF_DATASET_NAME,
+        repo_type="dataset"
+    )
+    
+    os.remove(local_file)
+
+
+
+
+
+def submit_annotation(
+        user_id, 
+        age_group, 
+        gender, 
+        native_language, 
+        tts_experience, 
+        device_type, 
+        selected_model,
+        context, 
+        score,
+        context_score
+):
+   try:
+
+    submit_annotation_hf(
+        user_id, 
+        age_group, 
+        gender, 
+        native_language, 
+        tts_experience, 
+        device_type, 
+        selected_model,
+        context, 
+        score,
+        context_score
+    )
+
+    return f"Annotation saved for {context}."
+
+   except Exception as e:
+       print("Submit error")
+       print(type(e).__name__)
+       print(e)
+
+       return f"ERROR: {e}"
+

services/audio_service.py

@@ -0,0 +1,15 @@
+from data.contexts import CONTEXTS
+from models.manager import generate_audio
+
+def generate_audio_from_context(
+        model_name,
+        context
+):
+    sentence = CONTEXTS[context]
+
+    audio = generate_audio(
+        model_name,
+        sentence
+    )
+
+    return audio, {"played": 0}