Can You Not Hear Me Now??? A review of Voice Dial, the first native voice command app for the iPhone
A good voice dialing / voice control application is a must-have for lots of people. A voice command program can be a real killer app when you’re driving or in quite a few situations where simply saying a command – let’s just say Notebook for pulling up your Google Notebook page – is the fastest way to get something launched.
Until recently, there were no voice command apps available for the iPhone – and many of us who were used to using one on Palm or Windows Mobile devices were sorely missing having one around.
Then a few weeks back, Makayama launched Voice Dial, the first native voice command app for the iPhone, and lots of us were very glad to see it. Voice Dial is intended to be a paid-for application – something that is still fairly rare for apps like this that requite that you jailbreak your iPhone in order to use them. It will normally sell for $32.95, although it is being offered at a special price of $27.95 right now. That’s a relatively high price for a mobile application, so Voice Dial is obviously aiming to be considered a premium-level iPhone application.
Read on to see whether it’s living up to its ambitions …
I’ve been using and testing Voice Dial since the day it first came out, back on January 22nd. I’ve used the first 1.01 version and have been running the latest 1.02 version for all my final testing of the app in the last several days.
Here’s what I’ve found in working with Voice Dial …
Installation:
Voice Dial – like the vast majority of current 3rd party apps for the iPhone – is installed via the Installer.app program, and is listed under the Productivity category. If you don’t see the latest version (1.02) listed, you may need to add Makayama’s source to Installer, which you can do as follows:
- Open Installer and tap the Sources tab
- Tap the ‘Edit’ button at the top right
- Tao the ‘Add’ button at the top left
- In the address field, after http:// – enter tinyurl.com/2t8cax
Usage:
Voice Dial is well laid-out and easy to use. The app prompts you when you first launch it to setup the three basic system commands – Accept, Repeat, and Cancel – and will nag you until you’ve got those done. Once you’ve done those, you can go to the Actions tab in the program and start setting up your own custom voice commands.
The three system commands are used for you to provide confirmation (or not) once Voice Dial has tried to match a voice command you’ve given it – like so: Accept will perform the action. Repeat enables you to speak again after a false recognition. Cancel will stop the recognition process.
You can enter voice commands in three main areas – Contacts, B
ookmarks, and Applications. When you choose Contacts, once you select a contact, you need to also tap on an individual phone number entry (so you can select a person’s mobile or home number etc.) in order to proceed with voice tag recording – this threw me at first, as I thought just tapping the contact name would be sufficient.
In the Bookmarks section, you can browse through all of the bookmarks you have in Mobile Safari, and record voice tags for any of them.
Finally, the Applications section lets you scroll through and create voice tags for any of the applications installed on the iPhone – including any 3rd party apps, not just built-in ones.
I think it’s great that Makayama chose to include the ability to launch your favorite bookmarked sites and web apps. I don’t recall seeing this feature on other mobile voice command apps, and it’s a great one to include with the iPhone, given how heavily the Safari browser gets used.
Creating a voice tag is quick and easy. As soon as you tap on a contact phone number, a bookmark, or an application, you’re prompted to hit ‘OK’ when you’re ready to record. Once you hit OK, you’ll be voice prompted to start talking – the app will say ‘Speak Now’ to let you know when to start recording. And once you’ve spoken your command, it will repeat it back to you and offer ‘OK’ or ‘Cancel’ buttons to accept your recording, or cancel it and try again.
There is not much more to the program than just setting up your voice 
commands, and setting a few preferences. The preference settings you can make use of are:
- Set Voice Dial to launch via a double-tap of the home button – this can be toggled on and off
- Set whether or not you want voice confirmations each time you give it a command, and whether you want a voice guide to tell you when it is ready to accept commands after you launch it – if set to on, you’ll get the ‘Speak Now’ voice prompt as soon as Voice Dial is ready to take commands.
Since Makayama intends that Voice Dial will be a full paid-for app once the iPhone SDK is out, we currently use the ‘trial’ version of it. To remind you of the trial nature of the product, a ‘nagware’ element is used and prompts at every launch of the program to ‘Buy’ or ‘Continue’. This seems over the top to me, and not well implemented. In fact, it strikes me that this makes the whole experience less of a proper trial of the app – because one of the key things about a voice command app is that I want to be doing things by voice when interacting with it as far as possible – not touching the screen. The reminder prompt should be less frequent – every 5 or 10 launches of the app maybe.
Tips for Getting Better Results
I’ve definitely found that you get much better results if you ensure that you’re in a quiet place when you record your commands – no TV or music in the background etc.
Here are a few of the other tips straight from Makayama’s manual for Voice Dial:
When recording names: record the first name and the last name without a pause . Our speech recognition algorithm doesn’t like pauses between words.
Record multiple voice tags for each action. You may store up to 3 voice tags per action. More voice tags mean a better chance of a positive recognition. Go to Commands and choose an item from the list. Then select Add Voice Tag to record additional tags.
When recording multiple voice tags: apply slight variations in the recordings. Use a different pitch for the second voice tag and a different tempo for the third.
To see the full listing of tips on improving results, see the Voice Dial Quick Start Guide, HERE.
Performance & Accuracy
While Voice Dial is in a must-have sort of category for lots of folks, it’s also a pretty simple, single purpose program. I just want to be able to create voice commands easily, and use them effectively.
So I’ve done some casual testing of Voice Dial – inside in quiet rooms, while driving, while out walking with the dog – and also some final tests in a q
uiet room where I recorded its results.
Bottom line on all of my testing so far – it performs very badly. This is true not just in the noisier environments like while driving, but even in my living room with no TV or music on, and pretty much zero background noise.
Makayama claims that:
VoiceDial foriPhone is speaker-dependent and requires a short period of training. It may then successfully capture your speech with a very high accuracy. It can achieve 96% accuracy if operated under optimal conditions.
For my final bit of testing I did as follows:
- Picked 17 of my 20 created commands – a mixture of bookmarks and applications (as I didn’t want to be calling folks inadvertently during the tests)
- Spoke each of them 5 times to Voice Dial in a quiet room, with no TV, music or background noise, with the iPhone held only 6-12 inches away from my face
The Final Score: 44 out of 85 commands recognized successfully. Just barely above 50%, and well short of the 96% accuracy it’s said to be able to achieve under optimal conditions.
I don’t know how close my testing would be to optimal conditions, but just above 50% accuracy in a quiet room, with the iPhone held fairly close to me, using all commands that had been recorded following Makayama’s suggested methods, and all commands that had the maximum 3 voice tags associated with them – seems a terribly poor result.
Another major performance weakness I’ve seen with Voice Dial is a major lag between the time you launch it and the time it is ready to accept commands. I have timed this quite a few times over the last couple of days, and it now consistently takes 10 seconds to be ready to accept commands. The program’s splash screen pops up quite quickly, but sits at the ‘Please wait – loading sound samples’ stage for most of the 10 seconds.
10 seconds feels very slow to me. It’s been a while, but I don’t recall Microsoft’s Voice Command having that sort of lag time.
The startup lag time grows as you add commands – and if my 20 commands already adds up to 10 seconds startup time, that does not bode well at all for anyone who plans to really hit this app with a large number of commands.
One other thing to mention in terms of testing out the application. In my final testing of the 17 commands, I had 3 spontaneous reboots of the iPhone. I run a large number of 3rd party apps, so there could be any number of causes for the reboots. But … since my last jailbreak via the Dev Team’s method (5 days ago) I have suffered no other spontaneous reboot while running any other application.
Conclusions
While it’s great to see a native voice command app for the iPhone, I am thoroughly unimpressed with Voice Dial thus far.
It’s easy to use and the range of commands it lets you setup are fine and well done – but its performance has been pretty awful for me. The lag time to be ready to accept commands is frustrating, the ‘nagware’ element of things is way too ‘in your face’, and the bottom line, the accuracy of its recognition, is very poor.
As with any app, and perhaps especially with a voice command program, your mileage may vary a whole lot – but I honestly felt the last few days like I wanted to turn into the Verizon TV ads guy, and just ask Voice Dial ‘Can You NOT Hear Me Now???”
I hope future versions of Voice Dial perform much better. They will need to I think to justify that fairly high price tag …
More Info:
For more information on Voice Dial you can visit its main page at Makayama’s site HERE.
UPDATE: Vincent at Makayama has provided a response to this review. Firstly, he has added a section to their FAQ page, as follows:
Less is more. If you store too many contacts, the recognition accuracy will drop, because you will have more people with similar sounding names, making it too difficult for the software to distinguish between them. To get 95% recognition rate, store the 10 phone numbers you dial most. You may store more, but the recognition rate will drop considerably.
He has also give some insight on the lag time / loading delay issue, as below:
About the loading delay:this will improve as soon as the SDK is available. We have to use a workaround now to store the sound samples, resulting in longer loading times, the more samples you store, the longer it takes.
Continue reading:
- Everything New Apple Just Announced (Septembe
- Apple Watch Pre-Order
- Apple Research Kit launches with 5 Apps
- Apple TV now only $69
TAGS:
