We are preparing to make some significant changes to our online recorder interface and general website. As of 8/3/19 we will be introducing the following changes and new features to simplify the workflow for selecting, recording, and finalizing inventories:
NOTE: Clients who have already started recording one of our current inventories (e.g., full, max) as of the go-live date will not see these changes and will be able to complete their current inventory.
New clients will not need to type an inventory name in the online recorder Settings dialog. Instead, the inventory name will either be preset, or selectable from a drop-down menu. The three inventory options will be listed as “Test Sentences”, “Custom Sentences”, and “Standard Sentences”
- To get started, a new client will need to complete 10 screening sentences as always. This inventory will be preset and will appear as “Test sentences”. This is exactly the same as the current “Screen” inventory and it will be the only inventory that clients can record until we have approved their hardware settings and advanced them to begin voice banking in earnest. As before, clients may need to redo this inventory if given specific feedback to correct any audio issues heard.
- After we have approved the Test sentences, clients will typically see a drop-down Inventory menu consisting of the next two inventories: “Custom sentences;” and “Standard sentences.” Clients may record in either of these inventories whenever they want and switch back and forth between them at will (though they will need to go through the Settings dialog again whenever switching). These two inventories consist of the following:
- Custom sentences: This inventory will have a few examples of standard message-banking sentences and personalizing sentences, but will primarily be present for people to add custom material (personal names, phrases, or words that are important to them) via our regular custom sentence form.
- Standard sentences: This inventory will contain all the predefined sentences from our previous “Max” inventory, ordered in such a way that the first sentences are the most important and successive sentences are less important to synthetic voice quality. There will be 3155 sentences in the list, but clients may choose to stop recording at any time they wish. Note that the number of sentences listed below are merely guidelines, differences in synthetic voice quality are often due more to the speaker’s natural voice quality, articulation, and speaking habits than to the number of sentences recorded. But we will continue to create both unit selection (USel) and hybrid DNN (hDNN) voices for clients to audition regardless of the number of sentences recorded.
Number of Sentences hDNN Voice Quality USel Voice Quality 400 Acceptable Questionable 800 Good Good 1600 Better Better 3155 Best Best
Finalizing or Requesting a voice
Since we will be giving all new clients the same inventory of 3155 possible sentences, most clients will probably want to stop before all the sentences have been recorded. To make it easier for clients to let us know when they have “had enough,” we will be unifying and simplifying the way clients finalize their recording. There will now be two Build My Voice selectors available, either one will allow the client to finalize their recordings. One will appear on our main webpage under the Voices menu as a menu option, and the second will appear as an actual button in the online recorder. The client will still need to login to activate the Build My Voice process, but will not need to select an inventory or go through the online recording settings to let us know they are finished recording. When Build My Voice is selected, all active inventories will be closed and finalized so that we can collect any recordings and begin to build voice versions for auditioning. This change will be visible to clients recording existing inventories as well as new clients.
Changes to the Main page Voices menu
In addition to the Build My Voice option, the Voices menu on our main page will contain new options for clients to audition donated voices and review any voices that the client has available for download, including their own personal voice and/or any donated voices they have requested access to download. We will gradually be introducing small fees for downloading donor voices. The fees will vary from voice to voice based on our costs in obtaining, building, and maintaining the voices. Clients who only wish to download a donated voice must still register with our system to access the donated voices. However, donated voices can be tested on our Demo page without registering.