Abstract: A method for performing zero-shot voice transfer using text-to-speech (TTS) includes receiving an input text sequence characterizing an utterance and receiving a reference speech representation characterizing a reference utterance spoken by a target speaker. The method also includes generating an encoded textual representation for the input text sequence, processing, using a speaker encoder, the reference speech representation to generate a speaker representation characterizing voice characteristics of the target speaker and learning fine-grained embedding vectors based on the speaker representation to obtain a final embedding vector. The method also includes predicting a duration and upsampling the encoded textual representation into an upsampled output. The method also includes generating a synthesized speech representation based on the upsampled output and the final embedding vector and generating a time-domain audio waveform of the input text sequence that clones a voice of the target speaker.

Abstract: A method includes receiving a reference utterance and an input text utterance. The reference utterance includes a plurality of terms spoken by a reference speaker and the input text sequence includes a corresponding transcript for each of the plurality of terms spoken by the reference speaker. The method includes obtaining a speaker embedding characterizing speaker characteristics of the reference speaker that spoke a plurality of terms. The method includes generating a replacement input text sequence by replacing the corresponding transcript of a respective one of the plurality of terms with a replacement transcript corresponding to a different term not included in the reference utterance. The method includes generating, using a text-to-speech (TTS) model conditioned on the reference utterance and the speaker embedding, resynthesized speech based on the replacement input text sequence in a voice of the reference speaker.

Abstract: A method includes receiving a set of training utterances each including a non-synthetic speech representation of a corresponding utterance, and for each training utterance, generating a corresponding synthetic speech representation by using a voice conversion model. The non-synthetic speech representation and the synthetic speech representation form a corresponding training utterance pair. At each of a plurality of output steps for each training utterance pair, the method also includes generating, for output by a speech recognition model, a first probability distribution over possible non-synthetic speech recognition hypotheses for the non-synthetic speech representation and a second probability distribution over possible synthetic speech recognition hypotheses for the synthetic speech representation.

Abstract: A method includes receiving a set of training utterances each including a non-synthetic speech representation of a corresponding utterance, and for each training utterance, generating a corresponding synthetic speech representation by using a voice conversion model. The non-synthetic speech representation and the synthetic speech representation form a corresponding training utterance pair. At each of a plurality of output steps for each training utterance pair, the method also includes generating, for output by a speech recognition model, a first probability distribution over possible non-synthetic speech recognition hypotheses for the non-synthetic speech representation and a second probability distribution over possible synthetic speech recognition hypotheses for the synthetic speech representation.

Abstract: A method includes receiving training data including transcribed speech utterances spoken in a general domain, modified speech utterances in a target domain, and unspoken textual utterances corresponding to the transcriptions of the modified speech utterances in the target domain. The modified speech utterances include utterances spoken in the target domain that have been modified to obfuscate one or more classes of sensitive information recited in the utterances. The method also includes generating a corresponding alignment output for each unspoken textual utterance of the received training data using an alignment model. The method also includes training a speech recognition model on the alignment outputs generated for the corresponding to the unspoken textual utterances, the un-transcribed speech utterances, and the transcribed speech utterances to teach the speech recognition model to learn to recognize speech in the target domain and phrases within the one or more classes of sensitive information.

Abstract: A method includes receiving a set of training utterances each including a non-synthetic speech representation of a corresponding utterance, and for each training utterance, generating a corresponding synthetic speech representation by using a voice conversion model. The non-synthetic speech representation and the synthetic speech representation form a corresponding training utterance pair. At each of a plurality of output steps for each training utterance pair, the method also includes generating, for output by a speech recognition model, a first probability distribution over possible non-synthetic speech recognition hypotheses for the non-synthetic speech representation and a second probability distribution over possible synthetic speech recognition hypotheses for the synthetic speech representation.

Abstract: A method includes receiving training data that includes unspoken text utterances, un-transcribed non-synthetic speech utterances, and transcribed non-synthetic speech utterances. Each unspoken text utterance is not paired with any corresponding spoken utterance of non-synthetic speech. Each un-transcribed non-synthetic speech utterance is not paired with a corresponding transcription. Each transcribed non-synthetic speech utterance is paired with a corresponding transcription. The method also includes generating a corresponding synthetic speech representation for each unspoken textual utterance of the received training data using a text-to-speech model. The method also includes pre-training an audio encoder on the synthetic speech representations generated for the unspoken textual utterances, the un-transcribed non-synthetic speech utterances, and the transcribed non-synthetic speech utterances to teach the audio encoder to jointly learn shared speech and text representations.

Abstract: A method includes receiving audio data corresponding to an utterance and generating a pair of positive audio data examples. Here, each positive audio data example includes a respective augmented copy of the received audio data. For each respective positive audio data example, the method includes generating a respective sequence of encoder outputs and projecting the respective sequence of encoder outputs for the positive data example into a contrastive loss space. The method also includes determining a L2 distance between each corresponding encoder output in the projected sequences of encoder outputs for the positive audio data examples and determining a per-utterance consistency loss by averaging the L2 distances. The method also includes generating corresponding speech recognition results for each respective positive audio data example. The method also includes updating parameters of the speech recognition model based on a respective supervised loss term and the per-utterance consistency loss.

Abstract: A water filter includes a housing, a spoiler unit, a power generation unit, a sterilization unit and a purification unit. The housing includes an input portion, an output portion and a first thread portion. The spoiler unit is provided in the housing and corresponds to the input portion. The power generation unit is provided in the housing and is coupled to the spoiler unit. The sterilization unit is provided in the housing and is electrically connected to the power generation unit, and includes a sterilization light source. The purification unit is provided in the housing and corresponds to the output portion and the sterilization unit, and includes a plurality of purification particles. Thus, the water filter is enabled to be coupled to a faucet using the first thread portion, allowing tap water to flow toward the direction from the input portion to the output portion.

Abstract: The invention relates to (S)-3?-methyl-abscisic acid, and esters thereof, and methods of using and making these compounds.

Abstract: The invention relates to (S)-3?-methyl-abscisic acid, and esters thereof, and methods of using and making these compounds.

Abstract: A radio signal receiving system for providing a signal to a transceiver includes a signal retrieving module and a signal processing module. The signal retrieving module retrieves a radio signal through one of a conducting wire in an electrical outlet, a conducting wire in a vehicular cigarette lighter, and a metallic vehicular casing. The radio signal receiving system operates without any conventional self-contained antenna and includes a radio signal receiving carrier which is either made from a conventional conducting wire or made of a metal to thereby enhance the efficiency of signal reception.

Abstract: A radio signal receiving system for providing a signal to a transceiver includes a signal retrieving module and a signal processing module. The signal retrieving module retrieves a radio signal through one of a conducting wire in an electrical outlet, a conducting wire in a vehicular cigarette lighter, and a metallic vehicular casing. The radio signal receiving system operates without any conventional self-contained antenna and includes a radio signal receiving carrier which is either made from a conventional conducting wire or made of a metal to thereby enhance the efficiency of signal reception.

Abstract: A mutually inductive resonant antenna receiving radio waves of dual frequency bands improves a conventional antenna series-connected to a uniaxial wire. The mutually inductive resonant antenna receives FM or TMC radio waves and comprises a first antenna and a second antenna. The first antenna has a first conductive core wire and a first insulating layer. The first insulating layer encloses the first conductive core wire. The second antenna has a second mesh-like conductive layer and a second insulating layer. The second mesh-like conductive layer encloses a section of the first antenna such that another section of the first antenna is exposed. The second insulating layer encloses the second mesh-like conductive layer. A section of the second mesh-like conductive layer is extended from the first antenna and electrically connected to a signal transmission line. The second mesh-like conductive layer is not in contact with the first conductive core wire.

Abstract: A mutually inductive resonant antenna receiving radio waves of dual frequency bands improves a conventional antenna series-connected to a uniaxial wire. The mutually inductive resonant antenna receives FM or TMC radio waves and comprises a first antenna and a second antenna. The first antenna has a first conductive core wire and a first insulating layer. The first insulating layer encloses the first conductive core wire. The second antenna has a second mesh-like conductive layer and a second insulating layer. The second mesh-like conductive layer encloses a section of the first antenna such that another section of the first antenna is exposed. The second insulating layer encloses the second mesh-like conductive layer. A section of the second mesh-like conductive layer is extended from the first antenna and electrically connected to a signal transmission line. The second mesh-like conductive layer is not in contact with the first conductive core wire.

Abstract: A digital broadcasting antenna structure includes a substrate having at least a first and a second face; a main antenna arranged on the first face; an amplifier arranged on the first face and electrically connected to the main antenna; a compensating unit arranged on the second face and electrically connected to the main antenna; a bandwidth modulating unit arranged on the second face and electrically connected to the compensating unit; and a grounding section arranged on the second face and electrically connected to the bandwidth modulating unit. The digital broadcasting antenna structure can receive digital broadcasting signals without being restricted to any specific receiving direction, and is applicable to low, intermediate and high frequency bands to therefore achieve the effects of miniaturization, high bandwidth and low return loss.

Abstract: A wireless signal conversion system includes a conversion output apparatus and at least one conversion input apparatus. The conversion output apparatus receives a wireless signal via an antenna, and feeds signal data carried by the wireless signal into a power line. The conversion input apparatus retrieves the signal data from the power line and then provides the signal data to an electronic device.

Abstract: A keypad is used to enter complex characters using a phonetic input method editor (IME). The user may enter complex characters by combining consonants, vowels, mid-vowels and tones by selecting keys on a the keypad instead of using a full size keyboard. Instead of a one-to-one mapping between the symbols and keys on a full size keyboard, multiple symbols are assigned to single keys on the keypad. For example, on a keypad having ten keys an average of four phonetic symbols are mapped to each of the ten keys on the keypad. The phonetic symbols are applied to the keypad in layers. For example, the symbols may be may be mapped to a consonant layer; a middle vowels+vowels layer; a vowels layer and a tone layer. Phonetic symbols with similar readings may also be mapped to the same key.