^^^^ That's what I was thinking, too. Slowing down by 50% hardly takes place without artefacts of several kind knowing that the original medium is analog tape.
Speeding up might produce a better result. Well, I've never tried so maybe my pessimism is wrong. 
 
As some others, I wonder, what was the reason to digitize the tapes at wrong speed. 
 
 

Kalle RantaahoAs some others, I wonder, what was the reason to digitize the tapes at wrong speed. 

Lazy digitising without listening to the tapes? Only a single speed tape deck available?

If the tapes are an assortment recorded at say 7.5ips, 15ips and 30ips but the person doing the digitising just ran them all through the tape deck at 15ips that would explain it.

Whatever the reason it's a far from trivial job trying to sort that out.

If the tape is recorded at half or double speed and the length changes as a result, you can use the Loop Construction window (but not to create a groove clip per se) to shift pitch and correct length at the same time.

Thanks guys, I was really interested to read that I hadn't missed something simple.  I think metelus and others have found the path to follow outside SONAR. 
 
The issue is that some tape is 2 track, some 4 some mono, some stereo.  Some is recorded at 15ips, some at 7, some 3.5 and - just possibly - half that.  I don't have all the playback machines with all the speeds.
 
Also some of the tape is very old (1960s).  Some of it is studio, some just mics left hanging out recording live at gigs, some it is demos.   Often you only get one real shot at threading to find out what you have.  
 
But I know that recording and playing back with a Roland RAP-10 card in the 90s worked just fine.  That Roland card was bundled with ATW (Audio Tool Works): all you had to do was tell ATW to play at double the sample rate in an simple varispeed playback.
 
So I think that Audacity (which I've used) could work just the same.  Great suggestion - I will report back shortly!

Some more info for you (edited out out of context info):
(from http://forums.stevehoffman.tv/threads/software-for-changing-tape-speed.334356/#post-9641432 )

The only true tape equalization difference is between NAB (15 IPS) and CCIR (30 IPS),..
The equalization problem occurs here because you're either halving or doubling the NAB curve, moving it either up or down an octave. This was a real problem with some CBS half-speed mastered albums, when they didn't re-equalize the playback equipment to compensate. So when you double the 7.5 IPS tapes to 15 IPS, it's going to have too much top end and not enough bottom end. This picture is what the NAB curve looks like. ..
The easiest thing to do would be to just apply a filter with a bit of "tilt" to the new file, to slightly reduce the highs and boost the bass. It actually sounds a lot more complicated than it is, and the resulting file should be pretty decent. 

Attached Files:
NAB curve.jpgFile size:51.5 KB
 

Information on manually changing sample rate by editing wav file:
If you don't have Audition or RX and Audition or whatever other software you have doesn't allow this kind of direct change of the sample rate stored in the file, it's possible to change the file using a freeware hex editor called Neo (http://www.hhdsoftware.com/free-hex-editor).

This takes a bit more careful attention. The following site explains how wave files headers are structured: https://ccrma.stanford.edu/courses/422/projects/WaveFormat/

Here's the example posted on that site with the four numbers represented sample rate larger and bolded:


52 49 46 46 24 08 00 00 57 41 56 45 66 6d 74 20 10 00 00 00 01 00 02 00
22 56 00 00 88 58 01 00 04 00 10 00 64 61 74 61 00 08 00 00 00 00 00 00
24 17 1e f3 3c 13 3c 14 16 f9 18 f9 34 e7 23 a6 3c f2 24 f2 11 ce 1a 0d​

Those numbers represent in this case, a sample rate of 88,200 but as written in hexadecimal format. If you switch the Windows calculator to Scientific in the View menu, choose the Dec format button under the number display, enter 88200 then change to Hex format. You will get 15888. the way a wave file header stores this information is by groups of two digits reversed so 15888 becomes 88 58 01 00 (with zeros padding out to four groups of two digits).

Here are the common sample rates plus any you'll use for doubling or halving

22050: 22 56 00 00
24000: C0 5D 00 00
44100: 44 AC 00 00
48000: 80 BB 00 00
88200: 88 58 01 00
96000: 00 77 01 00
176400: 10 B1 02 00
192000: 00 EE 02 00
352800: 20 62 05 00
352800: 20 62 05 00
384000: 00 DC 05 00
​
I know this might seem complicated but once you do it a couple of times, the procedure is really quite simple. The Neo hex editor I mentioned above is easy to use and provides the possibility of backing up the file you're working on in case you get it wrong. Practice a couple of times and you'll be fine and the results will be better sounding than speed changing software.

As for freeware resampling that does a decent job, there are several, but one of the best is called SoX (http://sox.sourceforge.net/). Note that it's a command line (DOS) program, so it takes getting used to. You could also use the freeware program Foobar 2000 (http://www.foobar2000.org/) and then install an add-on to it for the SoX resampler (http://www.hydrogenaudio.org/forums/index.php?showtopic=67373).
 
Source: http://forums.stevehoffma...d.334356/#post-9641978
 

I am waaaay post tape/analog era audio nerd but it may be helpful to the older doods (I don't mean that as perjorative... perhaps more experienced with tape would be more appropriate) if you said what playback devices you do have available to you. They may be able to recommended ways to get the actual tape devices to play back at the proper (or at least as close to proper) speed.
 
The idea when transferring to digital like this is to get the best signal from the tape outputting from the analog gear into the computer. In this case of course ideally you want the tape to playback at the correct speed but if that is impossible then using the CLOSEST speed and capturing that would mean less digital manipulation after the fact (which would result in less potential digital artifacts). Also the REALLY experienced dudes may be able to give you some tricks to fiddle with the tape machines in some wizardly way to force them to playback at a more appropriate speed. Like "oh yeah... I know that machine. It has whatchamagigdget setting. Twist it to that then do this in the computer" or "If you yank off panel A you'll see Screw B and you can adjust playback speed".
 
But yeah... get the playback speed from the device as close as you can going into the box and it will result in better quality when you bend it in the digital environment. Personally if I had sessions that were really important I would go so far as to even rent the proper tape device(s) for the digitization process.
 
Cheers and cool topic.

Try this:
insert a software synth capable of playing wav files (Drop Zone, Dimension, Rapture, SFZ player);
drag your file into the synth;
now you can vary the speed+pitch of the file by "playing" it on different keys. Let's say the original pitch is on C3, in that case C2 will play one octave lower.
Now insert long C2 note into midi track. The note suppose to be at least twice as long as the file, so it will play the file to the end.
Freeze the synth. 

Interesting.  I don't have anything to suggest to try and resolve the problems.  But, I come from the tape era and maintained a studio in Kansas City, Mo in the 70's (Big K Records) that had a 4 track Teac recorder and a 2 track Ampex.