Amazon Technologies, Inc.

Jonathan Barnett Feinstein

Brandon Scott Durham

Edward C Bueche

Toby Ray Latin-Stoermer

Fabian Moerchen

Panagiota Karanasou

Alok Verma

Catherine Breslin

Fabian Triefenbach

Amina Shabbeer

Judith Gaspers

Klaus Reiter

Kintan Dilipkumar Brahmbhatt

Rickesh Pal

Daniel Paul Ryan

Yun Suk Paik

Techniques are provided for training a target language model based at least in part on data associated with a reference language model. For example, language data utilized to train an English language model may be translated and provided as training data to train a German language model to recognize utterances provided in German. By utilizing the techniques herein, the efficiency of training a new language model may be improved due at least in part to replacing labor-intensive operations conventionally performed by specialized personnel with machine-generated data. Additionally, techniques discussed herein provide for reducing the time required for training a new language model by leveraging information associated with utterances of one language to train the new language model associated with a different language.

Techniques for model training for voice features

Techniques are provided for training a language recognition model. For example, a language recognition model may be maintained and associated with a reference language (e.g., English). The language recognition model may be configured to accept as input an utterance in the reference language and to identify a feature to be executed in response to receiving the utterance. New language data (e.g., other utterances) provided in a different language (e.g., German) may be obtained. This new language data may be translated to English and utilized to retrain the model to recognize reference language data as well as language data translated to the reference language. Subsequent utterances (e.g., English utterances, or German utterances translated to English) may be provided to the updated model and a feature may be identified. One or more instructions may be sent to a user device to execute a set of instructions associated with the feature.

Approaches provide for using a voice communications device to control, refine, or otherwise manage the playback of media content in response to a spoken instruction. For example, the voice communications device can receive a request to refine and/or initiate the playback of media content, such as music, news, audio books, audio broadcasts, and other such content. Audio input data that includes the request can be received by the voice communications device and an application executing on the voice communications device or otherwise in communication with the voice communications device can analyze the audio input data to determine how to carry out the request. The application can determine whether there is an active play queue of media content configured to play using the voice communications device. In the situation where there is no media content being played using the voice communications device, the application can determine media content using information in the request. In the situation where there is an active play queue of media content, the information can be used to refine the play queue. Thereafter, the application can cause the media content associated with the active play queue to play using the voice-enable communications device.

Refining media playback

Disclosed are various embodiments for identifying a message deliverability problem. Responses are received from one or more client devices that include information that identifies whether a respective response is associated with a first group of messages or a second group of messages. A message deliverability problem for at least one of the first group of messages or the second group of messages may be identified based at least in part on the information included in at least a portion of the plurality of responses.

Identifying message deliverability problems using grouped message characteristics

Methods, systems and computer program products are disclosed for identifying a message deliverability problem. In an example system, a source email server divides a plurality of email messages intended for delivery to destination devices using at least two source IP addresses into at least two groups of messages. The source email server outputs each email message in one of the groups through a same one of the source IP addresses and outputs each email message in another one of the groups through a different one of the source IP addresses. Each of the email messages is configured to trigger a response that includes the source IP address of the email message in response to a user opening the email message and/or selecting a hyperlink within the email message. A deliverability analysis module identifies an email message deliverability problem associated with one of the source IP addresses in response to a comparison of the responses across the source IP addresses.

Growing community of inventors

Sebastopol, CA, United States of America

Jonathan Barnett Feinstein