Thanks a lot fiverr member

[uk1000]

Competition is getting stronger each day @gwyneth_galvin I just got this in my email and Google must think it will interest me:

image1184×734 172 KB

image1182×338 50.6 KB

Do you think synthesys is what the gig voice over was done in or something else (different text to speech?)? The first video about that sounds quite real though it sounds a bit robotic in parts (eg. towards the end).

edit: I assume it wasn’t using the above software given what the first few posts said about it being a basic one. The software you mentioned does sound like one of the better ones and there are a few different voices in it. Though it doesn’t seem to let you adjust things in the voice over it creates (like changing how a particular word is spoken).

I wonder what the 1% of non-human voice is?

Maybe the 1% was the bit the software created. 🙂

I wonder at what point does an audio file be become computer generated? eg. is an audio file that’s been edited still 100% non-computer generated/100% natural? What if software puts a pre-recorded sentence (of a human voice) into an audio file (after that sentence was typed in as text) - technically it would be computer generated audio but the result would be the same as the original recording of that sentence.

[Guest macbluezell]

Did your gig get deleted?

No,

Now, I have made changes. Like we’re using Google api and IBM Watson tools used.

Now my gig under in pending approval review mode.

[uk1000]

No,

Now, I have made changes. Like we’re using Google api and IBM Watson tools used.

Now my gig under in pending approval review mode.

I think it makes it clear enough in the gig description and gig image what it is (I didn’t see it previously) since it says what you are using to create it. Though maybe the title could be clearer (eg. maybe it shouldn’t say “record” in the title and maybe it should say it’s software/computer generated in the title).

Maybe instead of “99% human voice” it could say something like "human-sounding synthesised voice " or “human-sounding computer generated voice” or something - in case the percentage isn’t totally accurate. Or you could take the line out and maybe say to listen to your sample for an example of the quality/how the voice over will sound.

[Guest macbluezell]

I think it makes it clear enough in the gig description and gig image what it is (I didn’t see it previously) since it says what you are using to create it. Though maybe the title could be clearer (eg. maybe it shouldn’t say “record” in the title and maybe it should say it’s software/computer generated in the title).

Maybe instead of “99% human voice” it could say something like "human-sounding synthesised voice " or “human-sounding computer generated voice” or something - in case the percentage isn’t totally accurate. Or you could take the line out and maybe say to listen to your sample for an example of the quality/how the voice over will sound.

Thanks for your suggestions.

[]

Please check it again.

Thanks

[uk1000]

Thanks for your suggestions.

[https://www.fiverr.com/share/438PqR]

Please check it again.

Thanks

Yes it’s changed and seems better. I still think the gig title should probably say it’s computer generated though. The gig description says “I will record male voice-over narration” - it still says record and maybe it could still mention computer generated there.

For the voice over, is there anything that can be done to make it sound cleaner/clearer. Like around the 58 second mark, is there any clipping there (or some other artefact) and if so do you have any way to prevent that?

[Guest macbluezell]

Yes it’s changed and seems better. I still think the gig title should probably say it’s computer generated though. The gig description says “I will record male voice-over narration” - it still says record and maybe it could still mention computer generated there.

For the voice over, is there anything that can be done to make it sound cleaner/clearer. Like around the 58 second mark, is there any clipping there (or some other artefact) and if so do you have any way to prevent that?

Hi,

Thanks

[uk1000]

Hi,

Please check it again.

Thanks

It seems better now, including the title and description (ie. more clearly describes what is offered - I’m not sure “record” is the right/best word for it but it should be okay). The voice over sample doesn’t have the artefacts that the previous one did. Is the sample now a female voice over? -If so it now doesn’t match the part of the gig that says “Gender - Male” (I assume you couldn’t get the male one working without those artefacts? eg. around the 58 sec mark).

It’s good. The demo could sound more realistic/human-like though. But I assume all the more realistic ones cost a lot more (like the Sythesys one) so maybe it wouldn’t be profitable (enough) with more expensive text to speech software.

[Guest cubittaudio]

I don’t think we’re listening to the same demo, because this to my ears sounds like something made on a computer built in the 1980s.

I could type the text into Google Translate, record the audio, and it would sound more natural.

I know I’m probably coming across as harsh… I just don’t understand why anyone would pay money for this, when even a 20 year old computer could do a better job using the built-in voice to text tool?

[uk1000]

I don’t think we’re listening to the same demo, because this to my ears sounds like something made on a computer built in the 1980s.

I could type the text into Google Translate, record the audio, and it would sound more natural.

I know I’m probably coming across as harsh… I just don’t understand why anyone would pay money for this, when even a 20 year old computer could do a better job using the built-in voice to text tool?

I just don’t understand why anyone would pay money for this

I agree there is higher quality text to speech software but they seem more expensive and maybe it would be too expensive for the OP/not profitable enough. eg. the Synthesys software costs $497 to be able to record an unlimited amount of text. I think the Pro version (to change pronunciation of words) costs $67 extra. So at the price the OP is charging, including commercial use, around 70 purchases would be needed just to break even if that software was used (assuming they were all the default number of words). I suppose a cheaper (limited number of characters/words) option could be used if they went with similar software.

So maybe some people would buy it (assuming they were satisfied with the quality - which some may not be) just because they don’t know how to record the text to speech themselves (or maybe there are rights issues with doing that for what they want it for) or maybe they think it’s simpler for the OP to do it and because for the amount of words they need+commercial use it’s cheaper than most/all real voice overs (though if they wanted less words or didn’t need commercial use it might not be). But I do think if it was more human-like more people would be interested in buying it.

Also the OP says he’s using IBM Watson for it, so I assume that’s better quality than the most basic/oldest text to speech software (even though there are better ones than that).

[uk1000]

Maybe using high quality text to speech software with the voice over artist’s own voice (eg. words/phrases/ sentences they record for it) could help voice over artists if they ever get problems with their voice (eg. voice strain) - by requiring less of the script to be recorded, assuming what was done was specified in the gig description.