Please wait until the page has a next to every sample.
Before commencing the test please read the instructions carefully.
This test takes approximately 20 minutes to complete.
There are a total of 1.8mb of sounds to download and it may take a few minutes for the sound files to load. We have to load them all in advance as an interruption to the play back would spoil the test.
When human actors read out loud they try their hardest to sound spontaneous, as though they are speaking to us and not reading from a script. We want to find out if machines can do this, or whether it is a special acting skill reserved for humans.
- Each sample lasts about 40 seconds. Please go through samples 1, 2 3 4 5 and 6 in turn. Scroll down if you cannot see all 6.
- Play each sound one after the other in numerical order by clicking on the play button. The button will turn red as the sound starts to playback.
- There are 6 different samples. They all use the same text and they are all have the same voice quality. You will be asked to choose one 'best voice actor' and one ‘worst voice actor.’
- Do not complete the judgment column until you have heard all the samples once. You can use the Comment column to remind yourself of your first impression, but we would also like to examine your comments, so it would be good if you could write a comment for every sample. Simple, brief but legible notes will suffice.
- For the best actor select ‘best’ in the judgment column. For the worst actor select ‘worst’ in the judgment column.
- It would help us if you tell us what you think of each sample in the comment section. You may wish to point out features that stand out as particularly poor or comment on positive features.
- You may play the samples as often as you wish .
- You will need a quiet environment to listen to the recordings; the changes are sometimes very subtle.