Update pipeline tag to audio-to-audio

This PR updates the `pipeline_tag` for the `amphion/TaDiCodec` model from `text-to-speech` to `audio-to-audio`. This change more accurately reflects the model's core functionality as a speech tokenizer capable of speech reconstruction, enabling it to be found under the correct pipeline category at https://huggingface.co/models?pipeline_tag=audio-to-audio. The existing `Text-to-Speech` tag is retained in the general `tags` list to acknowledge the model's use in TTS applications.

README.md

@@ -6,12 +6,12 @@ language:
 - fr
 - de
 - ko
+library_name: transformers
 license: apache-2.0
-pipeline_tag: text-to-speech
+pipeline_tag: audio-to-audio
 tags:
 - Speech-Tokenizer
 - Text-to-Speech
-library_name: transformers
 ---
 
 # 🚀 TaDiCodec