Abstract: A speech model includes a sub-model corresponding to a vocal attribute. The speech model generates an output waveform using a sample model, which receives text data, and a conditioning model, which receives text metadata and produces a prosody output for use by the sample model. If, during training or runtime, a different vocal attribute is desired or needed, the sub-model is re-trained or switched to a different sub-model corresponding to the different vocal attribute.

Abstract: A speech model is trained using multi-task learning. A first task may correspond to how well predicted audio matches training audio; a second task may correspond to a metric of perceived audio quality. The speech model may include, during training, layers related to the second task that are discarded at runtime.

Abstract: A speech model includes a sub-model corresponding to a vocal attribute. The speech model generates an output waveform using a sample model, which receives text data, and a conditioning model, which receives text metadata and produces a prosody output for use by the sample model. If, during training or runtime, a different vocal attribute is desired or needed, the sub-model is re-trained or switched to a different sub-model corresponding to the different vocal attribute.

Abstract: A speech model includes a sub-model corresponding to a vocal attribute. The speech model generates an output waveform using a sample model, which receives text data, and a conditioning model, which receives text metadata and produces a prosody output for use by the sample model. If, during training or runtime, a different vocal attribute is desired or needed, the sub-model is re-trained or switched to a different sub-model corresponding to the different vocal attribute.

Abstract: During text-to-speech processing, audio data corresponding to a word part, word, or group of words is generated using a trained model and used by a unit selection engine to create output audio. The audio data is generated at least when an input word is unrecognized or when a cost of a unit selection is too high.

Abstract: A speech model is trained using multi-task learning. A first task may correspond to how well predicted audio matches training audio; a second task may correspond to a metric of perceived audio quality. The speech model may include, during training, layers related to the second task that are discarded at runtime.

Abstract: To prioritize the processing text-to-speech (TTS) tasks, a TTS system may determine, for each task, an amount of time prior to the task reaching underrun, that is the time before the synthesized speech output to a user catches up to the time since a TTS task was originated. The TTS system may also prioritize tasks to reduce the amount of time between when a user submits a TTS request and when results are delivered to the user. When prioritizing tasks, such as allocating resources to existing tasks or accepting new tasks, the TTS system may prioritize tasks with the lowest amount of time prior to underrun and/or tasks with the longest time prior to delivery of first results.

Abstract: To prioritize the processing text-to-speech (TTS) tasks, a TTS system may determine, for each task, an amount of time prior to the task reaching underrun, that is the time before the synthesized speech output to a user catches up to the time since a TTS task was originated. The TTS system may also prioritize tasks to reduce the amount of time between when a user submits a TTS request and when results are delivered to the user. When prioritizing tasks, such as allocating resources to existing tasks or accepting new tasks, the TTS system may prioritize tasks with the lowest amount of time prior to underrun and/or tasks with the longest time prior to delivery of first results.