fix(spk): 修复 speaker embedding 集群后的重新排序问题

msgk

2025-02-14 a8591060d3889cd7a72841fa32a7ee64b49db1d2

 fun_text_processing/text_normalization/en/verbalizers/verbalize.py

@@ -1,5 +1,3 @@


from fun_text_processing.text_normalization.en.graph_utils import GraphFst
from fun_text_processing.text_normalization.en.verbalizers.abbreviation import AbbreviationFst
from fun_text_processing.text_normalization.en.verbalizers.cardinal import CardinalFst
@@ -19,7 +17,7 @@
class VerbalizeFst(GraphFst):
    """
    Composes other verbalizer grammars.
    For deployment, this grammar will be compiled and exported to OpenFst Finate State Archiv (FAR) File. 
    For deployment, this grammar will be compiled and exported to OpenFst Finate State Archiv (FAR) File.
    More details to deployment at NeMo/tools/text_processing_deployment.

    Args:
@@ -39,7 +37,9 @@
        fraction_graph = fraction.fst
        telephone_graph = TelephoneFst(deterministic=deterministic).fst
        electronic_graph = ElectronicFst(deterministic=deterministic).fst
        measure = MeasureFst(decimal=decimal, cardinal=cardinal, fraction=fraction, deterministic=deterministic)
        measure = MeasureFst(
            decimal=decimal, cardinal=cardinal, fraction=fraction, deterministic=deterministic
        )
        measure_graph = measure.fst
        time_graph = TimeFst(deterministic=deterministic).fst
        date_graph = DateFst(ordinal=ordinal, deterministic=deterministic).fst