After try outs with my son’s primary school home movies and club day reports, I decided to start my own English wargame youtube channel in LWTV-style. But I have a big problem. Pronuncation. My non-native-accent.

I sound dumb. I speak Dunglish. Terrible.

I love English with a female French accent, it’s erotic; I love English with a strong Scottish accent, it’s James Bond; I love English with a strong Spanish accent, it’s the I know nothing me from Barcelona-Fawlty Towers accent; but I HATE my own strong Dutch accent.

I have club friends who are native speakers, and Dutch friends who sound very British. Producing vlogs, though, is a tedious and lonely job. I’d rather not want to bother a friend, ask him, “hey, come over to my house, I have a voice over text, to read aloud.”

So the heavily marketed 37$ Speechelo software (44$ in Holland) seemed a deal. It’s marketed as ‘AI enhanced robospeech’ with countless options.

Here’s the official video.

I bought the online software package (in fact a web account). Below, five short audio samples

• my own Dunglish (duh)
• robotic standard British English
• AI enhanced British English
• AI enhanced US English, ‘friendly’ variation
• AI enhanced Dutch

Value for money?

Hmmm… Maybe. It’s okay…ish.

The website says: “”We offer over 30 voices for you to choose from. You can preview each voice to hear and find the one that best fits your needs. Also, you can add breathing sounds, longer pauses in the speech and even choose the tone of the speech: Serious, Joyful or just normal.”

I was interested in the English voices. In the basic 37$ package I found several English voices, but most of them are standard robovoices, medium to low quality. Only a few AI enhanced voices were available, and only the US male voice ‘Billy Joe’ had the serious/joyful/normal style.

The Dutch female voice is crisp and clear, somewhat bitchy, but okay. Same quality as announcements in railway stations. No style variations.

In other words: Speechelo sells you a stripped down version of their package. 44$ is a relatively high price for a very limited choice of English AI-voices.

The interface is suboptimal. Input is sentence by sentence: enter a line, generate the voice, download the generated voice recording, add the next line, repeat. You can copypaste a longer text, but checking if the recording is good and correcting errors is clumsy if you have more than 2 sentences.

For my limited purposes the US male friendly voice is good enough. With some editing it can “narrate” parts of the action of my vlogs. Overall, though, the whole software package is less value for money than I hoped. The voices are useful, but I’m not overly enthusiastic. I wished I’d been a native speaker!