Update README.md

README.md

@@ -111,6 +111,12 @@ LibriTTS-R [1] is a sound quality improved version of the LibriTTS corpus
 (http://www.openslr.org/60/) which is a multi-speaker English corpus of approximately 
 585 hours of read English speech at 24kHz sampling rate, published in 2019.
 
+
+## Transcoding notes
+
+- Source dataset: `mythicinfinity/libritts_r`
+- Target codec: `opus` `48kbps`
+
 ## Overview
 
 This is the LibriTTS-R dataset, adapted for the `datasets` library.
@@ -232,11 +238,3 @@ load_dataset("blabble-io/libritts_r", streaming=True)
 
 
 
-## Transcoding notes
-
-This dataset was generated by transcoding every top-level Hugging Face `Audio` column from the source dataset into a new codec while preserving all discovered configs and splits.
-
-- Source dataset: `mythicinfinity/libritts_r`
-- Target codec: `opus`
-- Generated by: `hf_audio_transcode_reupload_v3.py`
-- Multiple configs may reference the same uploaded Parquet files when identical transformed outputs were detected.