Adde comment on deduplication

README.md

@@ -30,13 +30,12 @@ size_categories:
 
 ## Dataset Summary
 
-**Utterly** is a speech dataset derived from the *human_5_all* and *smart-turn-data-v3.1-train* by Pipecat-AI. It contains over **6k English utterances** by a broad range of speakers, and augments conversational audio with turn-level annotations, including:
+**Utterly** is a speech dataset derived from the *human_5_all* and *smart-turn-data-v3.1-train* by Pipecat-AI. It contains over **5.8k English utterances** by a variety of speakers, and augments each example with turn-level annotations, including:
 
 * Verbatim Whisper-generated transcripts
 * End-of-turn (EoT) markers
 * Speaker identifiers (Coming soon)
 
-
 The dataset is designed to support research and development of speech and dialogue systems that require joint modeling of **speech recognition** and **conversational turn-taking**, such as streaming ASR systems and real-time conversational agents.
 
 ---
@@ -47,10 +46,10 @@ The dataset is designed to support research and development of speech and dialog
 * **Language(s)**: English
 * **Modality**: Audio (speech; mono-channel; sampled at 16kHz), Text
 * **Interaction type**: Human conversational speech
-* **Utterances**: 3,860
+* **Utterances**: 5,860
 * **Speakers**: >100
 
-Dataset splits (e.g., train/validation/test) are not predefined and may be created by downstream users as needed.
+Dataset splits (e.g., train/validation/test) are not predefined and may be created by downstream users as needed. Care was taken to ensure examples are unique through deduplication of the underlying audio examples.
 
 Note that Utterly is a *derived dataset*. All audio originates from the base datasets, with additional annotations created by the dataset author.