dsRNA comes from annealing of self-complementary sequences

siRNA is ssRNA and it’s half-life should be shorter

Short non-coding RNA in the form of siRNA and miRNA (just spent a semester of learning this so see if I can explain it correctly):

MicroRNAs are the short hairpin RNA that you first talked about. These are usually in the genome where the active genes are, and has its own non-coding gene placement, that utilizes RNA pol II to produce transcription.
The transcript is called a primary miRNA, and it is several kilobases long (there is also polycistronic – called cluster miRNAs). This pri-miRNA is then processed by DGCR8/Pasha (grabs the bottom of the hairpin by the single stranded RNA that are not hairpin paired) and Drosha (that has the cleaving site) and forms the precursor microRNA (pre-miRNA).
This pre-miRNA is then transported thru the nuclear pore protein Exportin 5. Once in the cytoplasm, it is processed by Dicer (cleaving action) and depending on the animal, TRBP/TARBP2/LOQ2 grab the mature miRNA and load it into the RISC (RNA-induced silencing complex) with the Ago proteins grabbing it.
The mature 20-22 nts miRNA can be taken from the 5p end of the pre-miRNA, or the other side of the hairpin, the 3p end. Once cut, the 5′ end of the miRNA has a ‘seed’ sequence that binds to the mRNA (usually the 3’UTR, but there are exceptions of course – promoter regions and introns). With miRNAs, the binding of the whole 20-22 nts mature miRNA is not perfectly Watson-Crick paired to the mRNA (just the seed part), and this translates into just binding up the mRNA and not cleaving it right away (translational repression).

Now the short interferring RNAs (siRNAs) are not processed by Drosha/Pasha(DGCR8). These are not hairpin making genes. They are from inverted repeats in the whole transcript that will bind up to become double stranded DNA, or it is a transcript from the coding strand that then binds to the transcript from the noncoding strand that also produces the double stranded DNA.
These are then transported to the cytomplasm where Dicer and its associated proteins process it into a mature siRNA (usually a little longer than the miRNAs -21-25nts) and can be placed into a different Ago and into the RITS (RNA-induced transcriptional Silencing) complex.
Usually the siRNAs bind to the mRNA in a perfect Watson-Crick base pair, and the RITS complex cleaves and degrades the RNA right away (transcriptional silencing).

You can google some of the main proteins above and get some diagrams that show what I am trying to explain, in order to get a visual idea of what is going on

thanks, its very helpful…