Text To Speech
There are lots of “Text To Speech” softwares around. Some are advertised as alternatives to real narrators”. I just wondered how they really sound like…
When I make a video, one of the essentials is the narration. Since videos I usually make are promotions of either products, services or companies, without narrations, they just don’t work.
The thing is, though, the cost of recording narrations. Lately, I get a lot of “We want to make videos but don’t have a budget”. To explain something, narrations are very useful but if my clients don’t have an enough budget, what can we do? So, recording narration has become an issue at our office.
Let’s see what these softwares are like in Japanese.
As far as I know, the latest is “ToSpeakOnline” by TOSHIBA. The service has just launched and there isn’t much infomation but it seems that TOSHIBA are pushing this service for educational contents
、narration for advertising、Web text to speech、etc. You can listen to actual voice samples but they all sound like robots…
Adobe counterpart, Grass Valley is selling “声の職人2(voice master)”, though it was developed by AI. This one seems to be developed as the “narrator alternative”. Voice samples sound like, well, I don’t want to be the one who has to say “We can use this as a narration” to my cllients
However, what I thought might work is the Anime version of this software. With “VOICEROID+ 鷹の爪 吉田くん“, you can make a narration with the voice of pretty famous Anime characters. This sounds pretty good though I can not use it for any serious production…
Now, let’s see how things are in English.
The price for hiring English speaking narrators are going down dramatically though it is still 120 – 150% of Japanese narrators. On top of that, things get troublesome since most recording directors do not speak English here and there’s a need for translators on site.
AudioPal provides a service with which you can put narration onto your site or blog. It covers not only English but tons of other languages. It even covers Japanese but one has to input Japanese in alphabets, and that is not very good. It lets you play the narration with Adobe Flash.
As a test, I’ve made it say “Thank you for visiting tokyo flaneur. Please look around and if you like something, please come back again”.
First British English. Surprisingly, it sounds very human.
Now, it’s American. Phmm, not bad but sounds artificial.
I came across “NaturalReader” by Naturalsoft. I’m not sure whether it’s because of the compression or original recordings but the sample sounds pretty noisy.
The other one is “IVONA Text-to-Speech“. The sound is clear and it doesn’t sound like a robot too much. The problem I had were the voice sounds pretty moody and the gaps between words which don’t sound natural.
We are thinking of making a promotion video for our company. The thing is, we can’t afford to ask professional narrators to come and do the recording. This is why I’ve been researching on these softwares. At the moment, we don’t want to use any of the softwares. I guess we have to do with our own voices…
Leave a Reply
