Dictating software

I was thinking of installing some software that will type the words in emails for example and one that comes to mind is the Nuance Naturally Speaking software.

Now I don't want to pay more than I really have to as I shall probably just use it for emails.

So any suggestions would be greatly appreciated.

Thanks for the reply Eric I have looked at others and the Nuance certainly looks to be the better of them all so I might treat myself to a premium copy of the software, and like you say it certainly beats typing eh??

You could use the Speech Recognition app that is in Windows 10 already.
Speech Recognition - TenForumsTutorials

If you like having this capability but want a more powerful / reliable / accurate application, you could then search for a replacement more effectively given the background experience that you would then have built up.

Denis

Thanks Steve and Try I will check them out

I spent many years as specialist IT support within local government assisting people with specialist needs. During that time I was involved with Dragon NaturallySpeaking well before it became Nuance and when Microsoft first introduced speech recognition within Windows Vista. Microsoft made a big thing about speech recognition at the time (probably after seeing that OS X - now macOS - included it). The company even added speech recognition support to MS Office for a brief period.

Unfortunately, Vista's speech recognition was absolutely dire and no amount of voice training helped recognition rates achieve better accuracy than ~50%. Windows 7's speech recognition was a big improvement, helped by third-party/enthusiasts' development of the add-on suite of 'speech recognition macros' that had been released initially with Vista (which I still have somewhere).

I've no idea about WIndows 8.x as I avoided it like the plague but - fast forward several years - I tested Windows 10's built-in speech recognition when it first came out and was very pleasantly surprised to find that recognition accuracy had rocketed upwards to ~70% straight out of the box, i.e. without any voice training at all. Unfortunately it was obvious that Microsoft had lost nearly all interest in speech recognition and its Win 10 interface is clunky to say the least, almost unchanged from iterations in previous Windows versions.

Whilst MS speech recognition works - often surprisingly well if you're careful about your choice of headset - it does take some perserverance (i.e. voice training) and is not anywhere near as well developed or supported as macOS speech recognition nor Nuance's continual speech recognition development with Dragon NaturallySpeaking and its off-shoots. With perserverance and a decent headset I would not be surprised if you couldn't achieve recognition accuracy of 80-85% if not better quite easily after training it to your voice.

By comparison, ScanSoft's Dragon NaturallySpeaking (DNS) was dying a death back in 2005 before being acquired by Nuance. Since then it has gone from strength-to-strength over the years, particularly with the release of DNS 9, 10 and 11. Recognition accuracy out-of-the-box rose swiftly during these 3 releases, even more so with DNS 12. I used DNS 12 extensively and was testing DNS 13 when I retired so have no further experience. With DNS 13 I was able to achieve recognition accuracy of 90% out-of-the-box.

So, in summary, if all you need is to dictate emails then the built-in Windows speech recognition is probably going to meet your needs... but depends entirely on the quality of headset you use and the amount of time you spend doing voice training (including memorising voice commands). If you buy a cheap headset then you will regret it forever and be disappointed.

If, however, you become dependent on voice recognition then a Nuance Dragon product will offer far more flexibility and better recognition accuracy, especially if you stick to the headsets and/or voice recorders that have been tested by Nuance.

If you like the idea of recording your emails on the move then transcribing them from voice recorder (or iPhone) later then DNS really is your only option due to its support for mobile recording and later transcription.

Whilst DNS was, in the past, a very expensive product (its Professional and Medical versions still are), its Home versions have plummetted in price. If I was looking at speech recognition today I would research headsets tested with DNS, buy one then use it with Win 10's built-in speech recognition for 2-3 weeks of testing/voice training at least then decide whether to accept its recognition accuracy or buy DNS instead.

Note that Win 10's speech recognition puts very little emphasis on voice training and is little more than an input check of volume/quality levels on first use. By comparison, DNS not only carries out the same input/quality checks but pushes voice training to the forefront during initial enrolment and afterwards... and provides much-needed support for backup of the voice profiles that are created. (Believe me, losing months and months of voice training down to a corrupted voice profile is soul-destroying...)

Final note: If you want to be able to script speech recognition then there's no longer any choice. MS' speech recognition macros died a death long before Windows 10 so only DNS Professional now supports scripting. Just sit down before you check the price.

Hope this helps...

Thanks Rick yes mate I have watched a few YouTubes on the different speech recoginition stuff but your reply I have to say has made me wonder if it is all worth it as I will have to work out how I am going to rig the headset/mic.

For example as I run a sound card and the sound system has it's own headphone jack and I wouldn't want to lay out that much money if I failed to get it working properly and what you said about it all falling in a heap if it fails because of some hardware being sub standard and I would het to lose a lot of voice training too if I had to change.