files added

app.py

@@ -0,0 +1,41 @@
+import gradio as gr
+import torch
+from datasets import load_dataset
+from transformers import SpeechT5Processor, SpeechT5ForTextToSpeech, SpeechT5HifiGan
+import soundfile as sf
+
+# Load the fine-tuned model, processor, and vocoder
+model_name = "microsoft/speecht5_tts"
+processor = SpeechT5Processor.from_pretrained(model_name)
+model = SpeechT5ForTextToSpeech.from_pretrained("emirhanbilgic/speecht5_finetuned_emirhan_tr")
+vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan")
+
+# Load the Turkish dataset
+turkish_dataset = load_dataset("erenfazlioglu/turkishvoicedataset", split="train")
+
+# Get an example text and its corresponding audio
+example_item = turkish_dataset[0]
+example_text = example_item['text']
+example_audio = example_item['audio']['array']
+
+# Create speaker embedding from the example audio
+with torch.no_grad():
+    speaker_embeddings = model.get_speaker_embeddings(torch.tensor(example_audio).unsqueeze(0))
+
+def text_to_speech(text):
+    inputs = processor(text=text, return_tensors="pt")
+    speech = model.generate_speech(inputs["input_ids"], speaker_embeddings, vocoder=vocoder)
+    sf.write("output.wav", speech.numpy(), samplerate=16000)
+    return "output.wav"
+
+# Create Gradio interface
+iface = gr.Interface(
+    fn=text_to_speech,
+    inputs=gr.Textbox(label="Enter Turkish text to convert to speech", value=example_text),
+    outputs=gr.Audio(label="Generated Speech"),
+    title="Turkish SpeechT5 Text-to-Speech Demo",
+    description="Enter Turkish text and listen to the generated speech using the fine-tuned SpeechT5 model."
+)
+
+# Launch the demo
+iface.launch()

requirements.txt

@@ -0,0 +1,5 @@
+transformers
+datasets
+soundfile
+torch
+torchaudio