The ordinary automatic speech recognition(ASR) frameworks utilize the GMM-HMM for acoustic modeling and the n-gram for language modeling. In the course of the last decade, the deep feed-forward neural network (DFNN) has nearly replaced the GMM in acoustic modeling. The current ASR systems are predominantly dependent on the DFNN-HMM acoustic model and the n-gram language model (LM). Inferable from better long-termcontext displaying capacity, the recurrent neural network(RNN) based LMs have as of now been accounted for to yield lower perplexitythan the n-gram LMs. As of late a variation of RNN, the long-short term memory(LSTM) has been effectively investigated inacoustic modeling. Strangely, the assessment of an ASR systemem ploying both RNN-based acoustic and semantic demonstrating is yetto be accounted for. Further, we note that most of these advancementsare explored in the context of adults’ ASR only. Persuaded bythose works, in this paper we investigate LSTM-based acoustic modeling joined with RNN-based LM for children’s ASR.Our exploratory outcomes show that such consolidated RNN-based modeling is found viable in both coordinated and mismatched children's ASR tasks.