Voice banking, or voice preservation, refers to a procedure where an individual is recorded while they speak a large volume of phrases. Specialized software uses these recordings to create an artificial voice that resembles the individual's.
The artificial voice created by voice banking software can then be used in the future to synthesize speech that sounds like the person's actual voice. Such a voice is generally well-received by listeners, as it feels much more authentic than a generic voice that is built-in to the software solution.
The phrases used to create the artificial voice are specially chosen to capture the widest possible variety of sounds, known as phonemes. Several phrases are used so that each phoneme is encountered multiple times; this increases the quality of the sampling and reduces the impact of a mispronounced word during recording.
The quality of the recording is directly impacted by the individual's ability to speak; in the case of ALS, it is important to make the recordings early on before speech capability declines any further.
No open standards exist for voice banking data formats. All voice banking data remains proprietary and locked into the software solution that created it.
For example, voice banking on macOS or iOS would not be of any use on Microsoft Windows or Android platforms.
Personal Voice is available on iPhone, iPad, and macOS PCs. Older hardware devices do not have Personal Voice available. The user must read 150 phrases aloud to create the voice, which requires a few hours of post-processing to complete. In the case of the English language corpus, many of these phrases are related to United States history.
✗ There are no voice banking solutions for Windows-based systems known to ALSWiki.org.
The following organizations offer voice banking services:
Message banking refers to making recordings of statements such as “I love you”, birthday wishes, and so forth. These are then saved for future listening by loved ones.