speech-edit
SpeechEditBench
SpeechEditBench
SpeechEditBench is a bilingual multi-attribute benchmark for
instruction-guided speech editing. Each example provides source speech and a
natural-language editing instruction, and the benchmark evaluates whether a
system can apply the requested edit while preserving the expected lexical
content.
Paper: SpeechEditBench: A Bilingual Multi-Attribute Benchmark for Instruction-Guided Speech Editing
Code and evaluator: github.com/daxintan-cuhk/SpeechEditBench… See the full description on the dataset page: https://huggingface.co/datasets/DiscreteSpeech/SpeechEditBench.speech_edit_acoustic
SpeechEdit Acoustic Retrieval Dataset
This dataset is an MTEB-formatted Any-to-Any (AT2A) composed audio retrieval adaptation of the acoustic_editing subset of DiscreteSpeech/SpeechEditBench.
Each query combines an original/source speech recording with a natural-language editing instruction, and the corpus contains the corresponding edited target speech recordings.
Schema
queries: id (string), audio (source audio), and text (edit instruction)
corpus: id (string)… See the full description on the dataset page: https://huggingface.co/datasets/deep9539/speech_edit_acoustic.droidnexus-arabic-editorial-speech-scorecard-mini
DroidNexus Arabic Editorial Speech Scorecard Mini
A public DroidNexus Labs scorecard dataset for Arabic speech workflows: representative editorial scenarios, latency targets, overlap pressure, and the metric stack that decides whether a transcript is usable.
Why this exists
This dataset is the first public speech artifact layer for DroidNexus Labs. It publishes representative editorial workloads and evaluation pressure before claiming a full source-audio benchmark.… See the full description on the dataset page: https://huggingface.co/datasets/driodnexus/droidnexus-arabic-editorial-speech-scorecard-mini.speech-edit-dataset这是一个用于语音编辑任务的测试集，涵盖中、英文两种语言，
涉及三种编辑任务：增加（insertion）、删除（deletion）和替换（substitution）。
每个样本都包含原始音频、音频路径、描述编辑操作的自然语言指令、原始文本转录，以及编辑后的目标文本转录。Speech-Editing-TestDemo-Speech-Editing