LOBE is a recording client made specifically for TTS data collections. It supports multiple collections, single and multi-speaker, and can prompt sentences based on phonetic coverage.
I am currently using tiro software to do it since the audio is amazing so I can just use ASR to "create the prompts."
The rights are a donation for kraft.is. How does one import a dataset into LOBE? is that possible? or if it is not currently possible, what is the TTS dataset creation guidelines and rules that were used for Talrómur? I believe these datasets were originally made within LOBE itself. Can you point me to the exact file which would be a good starting point?
If we ask nicely, the donor/donator/donuter :dango: might be willing to license his voice specifically for Icelandic language technology researchers within Iceland to use or even just for @atliSig to use to publish papers.
Hello. @atliSig @G-Thor @thdg
I have obtained the rights to create a TTS voice from this: https://www.youtube.com/watch?v=bkmfWA9hyPw
I am currently using tiro software to do it since the audio is amazing so I can just use ASR to "create the prompts."
The rights are a donation for kraft.is. How does one import a dataset into LOBE? is that possible? or if it is not currently possible, what is the TTS dataset creation guidelines and rules that were used for Talrómur? I believe these datasets were originally made within LOBE itself. Can you point me to the exact file which would be a good starting point?
If we ask nicely, the donor/donator/donuter :dango: might be willing to license his voice specifically for Icelandic language technology researchers within Iceland to use or even just for @atliSig to use to publish papers.