There is this output i get:

Duration: 00:03:27.05, start: 0.000000, bitrate: 336 kb/s

I want to extract 00:03:27.05 from this whole string(Values Can Change).

So far I have tried:

String="Duration: 00:03:27.05, start: 0.000000, bitrate: 336 kb/s"

echo $String

It gives the desired results, but here I have to Declare String variable Two times and have to cut specific part of string separately.

So I Just want to Know an easiest and Straight-Forward approach to do it.

Thanks in Advance for Help :)

  • Have you tried a regex based approach?
    – mtak
    Jul 21, 2022 at 12:19
  • Is it always in the same format of "Duration: NN:NN:NN.NN, Start: "... ?
    – Journeyman Geek
    Jul 21, 2022 at 12:20
  • 1
    This is the Right Way because the shell does it (no extra process(es) needed). But (1) quote, even while echoing. (2) String="${String%%,*}" may be better (see the citation here). Jul 21, 2022 at 12:25
  • yes! The Format remains the Same.
    – Random Guy
    Jul 21, 2022 at 12:25

3 Answers 3


You can try something like with awk:

awk -F'[ ,]' '{print $2}' input_file


echo "Duration: 00:03:27.05, start: 0.000000, bitrate: 336 kb/s"|awk -F'[ ,]' '{print $2}'
  • 1
    As Kamil Maciorowski noted, it's theoretically faster to do the processing inside the shell. But using awk is much clearer and easier to understand -- and change if needed. Bending over backwards to be fractionally faster is false economy here. If you really need that performance boost, don't do this in an interpreted scripting language at all.
    – mpez0
    Jul 21, 2022 at 13:02
  • 1
    @mpez0, awk will show full speed on more complex operations, this is just read, separate and print. And for me do not worth the effort to write it in C for example :) Jul 21, 2022 at 13:10
  • 1
    I'm trying to agree with your post and show that the convoluted do-it-all-in-bash isn't worth the hassle. That's why I upvoted it.
    – mpez0
    Jul 21, 2022 at 13:48

Assuming that this specific time pattern is the only one in your string, a very straightforward approach would be to simply use grep with the -P flag that allows us to interpret patterns as Perl-compatible regular expressions :

String="Duration: 00:03:27.05, start: 0.000000, bitrate: 336 kb/s"

echo $String | grep -oP "[0-9]{2}:[0-9]{2}:[0-9]{2}\.[0-9]{2}"

This would output 00:03:27.05


Bash lets you use POSIX ERE regular expressions:

String="Duration: 00:03:27.05, start: 0.000000, bitrate: 336 kb/s"

if [[ $String =~ Duration:\ ([^,]+), ]]; then

In plain Bourne shell, /bin/expr is the standard tool:

expr "$String" : 'Duration: \([^,]*\),'

(Note that expr uses BRE regex syntax instead of ERE, so grouping uses \( \) and the + operator is unavailable. Also, expr's regex is implicitly anchored to the beginning of the string, whereas =~ searches anywhere within the string.)

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .