🎤 SpeechT5 TTS Technical Train2

This model is a fine-tuned version of microsoft/speecht5_tts using a custom dataset, specifically trained for Text-to-Speech (TTS) tasks.

🎯 Key Metric:

• Loss on the evaluation set: 0.3763

📢 Listen to the generated sample:

The text is " Hello ,few technical terms i used while fine tuning are API and REST and CUDA and TTS."

---

📝 Model Description

The SpeechT5 TTS Technical Train2 is built on the SpeechT5 architecture and was fine-tuned for speech synthesis (TTS). The fine-tuning focused on improving the naturalness and clarity of the generated audio from text.

🛠 Base Model: Microsoft SpeechT5
📚 Dataset: Custom (specific details to be provided)

---

🔧 Intended Uses & Limitations

✅ Primary Use Cases:

• Text-to-Speech (TTS) for technical Interview Texts .
• Virtual Assistants:

⚠ Limitations:

• Best suited for English TTS tasks.
• Require further fine-tuning on Large dataset .

---

📅 Training Data

The model was fine-tuned on a custom dataset, curated for enhancing TTS outputs. This dataset consists of various types of text that help the model generate more natural speech, making it suitable for TTS applications.

⚙ Hyperparameters:

The model was trained with the following hyperparameters:

• Learning Rate: 1e-05
• Train Batch Size: 16
• Eval Batch Size: 8
• Seed: 42
• Gradient Accumulation Steps: 2
• Total Train Batch Size: 32
• Optimizer: AdamW (betas=(0.9, 0.999), epsilon=1e-08)
• LR Scheduler Type: Linear
• Warmup Steps: 50
• Training Steps: 500
• Mixed Precision Training: Native AMP

⚙ 📊 Training Results::

📦 Framework Versions

• Transformers: 4.46.0.dev0
• PyTorch: 2.4.1+cu121
• Datasets: 3.0.2
• Tokenizers: 0.20.1