fix cmn ASR eval in SeamlessExpressive #352
Open
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
SeamlessExpressive evaluation of eng-cmn normalizes both transcribed and ground truth texts into simplified Chinese texts. It is observed that there is a mix of traditional and simplified Chinese characters in texts, which resulted in low ASR-BLEU. Therefore, chinese_converter is added to ASR-BLEU in SeamlessExpressive eval pipeline.
Test command:
python3 ${SEAMLESS_COMMUNICATION_ROOT}/src/seamless_communication/cli/expressivity/evaluate/run_asr_bleu.py
--generation_dir_path=${GENERATED_DIR}/${SPLIT}${SRC_LANG}${TGT_LANG}
--generate_tsv_filename=${GENERATED_TSV}
--tgt_lang=${TGT_LANG}