SynthEmbed

A High-Fidelity Multimodal Document Query Engine

Models

1. Audio/Text segmentation

For Text--- https://github.com/notAI-tech/deepsegment https://medium.com/artiwise-nlp/text-segmentation-and-its-applications-to-aspect-based-sentiment-analysis-fb115f9ab4e9

For Audio---

2. Audio/Text Classification https://huggingface.co/superb/hubert-large-superb-er

3. Intent Classification https://huggingface.co/cartesinus/mdeberta-v3-base_amazon-massive_intent facebook/bart-large-mnli (best)

https://huggingface.co/models?pipeline_tag=text-classification&sort=trending https://huggingface.co/SamLowe/roberta-base-go_emotions

4. Speech to Text https://huggingface.co/facebook/s2t-large-librispeech-asr

5. Text to speech https://huggingface.co/microsoft/speecht5_tts

6. Embedding Model https://huggingface.co/TaylorAI/gte-tiny

7. LLMS https://huggingface.co/HuggingFaceH4/zephyr-7b-beta https://huggingface.co/mistralai/Mistral-7B-v0.1