creates text from audio of A/V input file, using docker, sphinx. extracts keywords and NLP entities. leverages OpenNews, Stanford, Oxford, CMU and more

Speak right into your todo.txt with speedo. Set tags, priority, due dates directly from speech.

Perform force alignment on Mandarin data using aidatatang pretrained model at https://kaldi-asr.org/models/m10

Resources for easily building ASR systems with Kaldi

Extract phone-level alignment and phonemic transcript from kaldi ali.*.gz files

Extract mfcc vectors and phones from TIMIT dataset

Hyfforddi a defnyddio modelau adnabod lleferydd Cymraeg coqui-stt a KenLM // Train and use coqui-stt and KenLM based Welsh language speech recognition models.

A Kaldi-Recipe for Egyptian Arabic Speech Recognition

An ASR recipe and speech corpus of Icelandic parliamentary speeches

Realtime internet radio stream speech recognition with Julius & ffmpeg

Scripts para treino de modelos acústicos

HHM-based Arabic ASR using Kaldi engine

A Kaldi recipe for training a hybrid DNN-HMM speech recognition model

A voice recorder and recording transmitter for Commbase

Open-set speech language identification https://arxiv.org/abs/2205.10397

Forced alignment of Nda‘ Nda’ a Cameroonian language

A voice recorder and recording transmitter for Commbase

Use kaldi pretrained nnet3 model to align individual sentences and get phone-level transcripts

Report of an end-to-end speech recognition task