DEAL Text to Speech Commercial License $27

[kenn10]

Haven't tried it but it looks exactly like Audacity.

 

[KNERD]

Haven't tried it but it looks exactly like Audacity.

View attachment 3862

Yes, it got forked after it the project was bought out by a commerical business and they started adding things like Telemetry.

 

[kenn10]

@krzykat I found this article on recording/converting recordings for use in Asterisk: https://support.voipcloud.online/hc...re-Upload-Voice-Prompts-to-PBX-Announcements-

Once you reduce a recording to the required sampling rate of 8,000, you lose a lot of the fidelity. I am experimenting with Audacity to reduce base and increase treble and volume then export it as a 16 KHz PCM wav file. I'm having some success but still can't get the lower sampling rate to sound as good as the original mp3.

 

[KNERD]

I wonder if there is a way to get around that 8k sampling rate Asterisk needs. I would assume cell phone comapines are switching over to VoLTE because of better voice quality. I mean what's the point of the 64k per channel of ulaw and alaw if the recordings are only 8k sampling.

 

[kenn10]

I wonder if there is a way to get around that 8k sampling rate Asterisk needs.

You'd think so since music on hold will accept an mp3 and gives decent fidelity. However, when I increase the sample rate for the recordings, Asterisk does not read them.

 

[KNERD]

You'd think so since music on hold will accept an mp3 and gives decent fidelity. However, when I increase the sample rate for the recordings, Asterisk does not read them.

Seems it does now

Home - Asterisk Documentation

wiki.asterisk.org

Looking over that URL, it seems you would have to start using Signed Linear, Speex, CELT, or SILK to get above the 8k sampling limit. It seems while ulaw, and alaw are 64k streams, they are sampled at 8kHz.

I wonder which providers offer those CODECS.


Looking at @Skyetel FAQ, they claim to support many CODECS

Because Skyetel does not sit in the audio path of the call, we will accept all codecs including OPUS, G722, etc. However, the PSTN only accepts G711.

If you have multiple PBXs on your account with us, you can configure your PBX to offer a higher quality codec with G711 as a fallback. This will allow your customers to talk to each other using G722 or OPUS while reverting to G711 when they call the PSTN. This is also true if your PBX calls another PBX on the Skyetel network. (in those cases we will attempt to bridge the audio directly from one PBX to another to increase the call quality.)

 

[MGD4me]

I mean what's the point of the 64k per channel of ulaw and alaw if the recordings are only 8k sampling.

8kHz is the sampling rate. Each sample is 8bits wide. 8000 samples X 8 data bits results in a 64k bit stream.

 

[krzykat]

FYI - I've gone in for the PRO version. If someone needs a sample of a voice on Pro, let me know. This product may cost a few bucks, but I think the results will be worth it.

 

[KNERD]

I see another company has upped the ante:

 

[kenn10]

Nice touches with scoff and breath, etc. Sadly, their pricing is based on usage, not a flat $47 like the other guys we've looked at. Currently, they only support American English and can emulate accents such as New York or Southern. This product is intended for large commercial entities. The do support API calls which would be great for voice response and text to speech in Asterisk but I suspect they'll be too rich for my blood when they announce pricing.

 

[krzykat]

Pretty nice, I like the idea of an API ... but what's the cost?

 

[kenn10]

Pretty nice, I like the idea of an API ... but what's the cost?

Thus far, undisclosed. See https://www.sonantic.io/

 

[KNERD]

It seems like they are targeting large multi-million dollar enterprises, but it could also be they trying to make it seem like they are a larger company than they really are.

I get the latter thought when I see things like this posted on company's web sites

 

[wardmundy]

Here's the script we're using with $27 SpeechVid to convert MP3 files to Asterisk GSM format:

#!//bin/bash

# script name:   mp3-to-gsm.sh
# script syntax: ./mp3-to-gsm.sh filename (without .mp3 extension)
# script yields: filename.gsm

if [ -z "$1" ]; then
 echo "Please append the name of the MP3 file to convert (without .mp3)"
 exit
fi

sox $1.mp3 --norm -r 8k -c 1 -e gsm $1.gsm

Using the English (U.S.) Penelope voice, it sounds pretty good to me after conversion.

 

[wardmundy]

There are almost 100 background music offerings included in the $27 license. Some would make excellent Music On Hold options, e.g. tropical-summer-beach-11171.mp3.

 

[JFrost]

It seems like they are targeting large multi-million dollar enterprises, but it could also be they trying to make it seem like they are a larger company than they really are.

I get the latter thought when I see things like this posted on company's web sites

View attachment 3958

This is a fairly standard block "element" in the current web design ethos.


Some of these evoke airs of a somewhat higher end country club membership...

I suspect they'll be too rich for my blood when they announce pricing.

I suspect you're right when the "Get Started" link wants me to contact their "team to find out how" they "can help" my "studio"

 

[wardmundy]

Here are samples of the .gsm conversions and translations from SpeechVid MP3's of the latest Nerd Vittles News Headlines.

The Spanish translation uses the $27 SpeechVid Worldwide (44-language translation) add-on.

 

[wardmundy]

WARNING WARNING WARNING

These are web-based tools. That means, if the seller's web site disappears, so does your product AND your money. You can protect yourself at least for six months by paying with PayPal which offers refunds if your product vanishes during the first six months after purchase. SpeechVid offers 30-day money-back guarantee.

WARNING WARNING WARNING

 

[KNERD]

Here are samples of the .gsm conversions and translations from SpeechVid MP3's of the latest Nerd Vittles News Headlines.

The Spanish translation uses the $27 SpeechVid Worldwide (44-language translation) add-on.

Sounds very robotic

 

[kenn10]

I still like Speechelo. I get very human sounding recordings from it and don't need music or a wide variety of non-English voices. Here are a couple of examples.