Update README.md

README.md

@@ -28,4 +28,8 @@ the encoder is based on nvidia's nano codec that encodes your audio to 22khz FSQ
 
 visit [dune_extraction.py](https://huggingface.co/Respair/dune_codec/blob/main/dune_extraction.py) <br><br>
 
-you need a dataset with audio embedded and a `key` column. if your metadata and audio are in one place, you can just DATASET_SOURCE == AUDIO_DATASET_SOURCE.
+you need a dataset with audio embedded and a `key` column. if your metadata and audio are in one place, you can just DATASET_SOURCE == AUDIO_DATASET_SOURCE.
+
+# important note
+
+this is strictly a speech tokenizer, trained only on human speech; it won't do well with music.