Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
import pynini
from fun_text_processing.inverse_text_normalization.vi.graph_utils import (
DAMO_NOT_QUOTE,
GraphFst,
delete_space,
)
from pynini.lib import pynutil
class ElectronicFst(GraphFst):
"""
Finite state transducer for verbalizing electronic
e.g. tokens { electronic { username: "cdf1" domain: "abc.edu" } } -> [email protected]
"""
def __init__(self):
super().__init__(name="electronic", kind="verbalize")
user_name = (
pynutil.delete("username:")
+ delete_space
+ pynutil.delete('"')
+ pynini.closure(DAMO_NOT_QUOTE, 1)
+ pynutil.delete('"')
)
domain = (
pynutil.delete("domain:")
+ delete_space
+ pynutil.delete('"')
+ pynini.closure(DAMO_NOT_QUOTE, 1)
+ pynutil.delete('"')
)
protocol = (
pynutil.delete("protocol:")
+ delete_space
+ pynutil.delete('"')
+ pynini.closure(DAMO_NOT_QUOTE, 1)
+ pynutil.delete('"')
)
graph = user_name + delete_space + pynutil.insert("@") + domain
graph |= protocol
delete_tokens = self.delete_tokens(graph)
self.fst = delete_tokens.optimize()