Tacotron
E746091
Tacotron is a neural network-based text-to-speech system that generates natural-sounding speech by predicting mel-spectrograms from text, often used in conjunction with neural vocoders like Parallel WaveNet.
All labels observed (3)
| Label | Occurrences |
|---|---|
| Tacotron canonical | 1 |
| Tacotron 2 | 1 |
| Tacotron: Towards End-to-End Speech Synthesis | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T8577279 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: Tacotron Context triple: [Parallel WaveNet, relatedTo, Tacotron]
-
A.
WaveRNN
WaveRNN is a neural network-based audio waveform generator designed as a more efficient, real-time alternative to earlier autoregressive models for tasks like text-to-speech synthesis.
-
B.
WaveNet
WaveNet is a deep generative neural network architecture for raw audio that produces highly natural-sounding speech and other audio signals.
-
C.
WaveGlow
WaveGlow is a flow-based generative neural network model for fast, high-quality text-to-speech audio synthesis.
-
D.
Parallel WaveNet
Parallel WaveNet is a neural vocoder architecture that accelerates high-fidelity audio waveform generation by distilling the autoregressive WaveNet model into a fast, parallelizable form.
-
E.
Wav2Vec2
Wav2Vec2 is a self-supervised deep learning model for automatic speech recognition that learns powerful audio representations directly from raw waveforms.
- F. None of above. chosen
- G. Unsure - the case is ambiguous/there is not enough information to decide.
NED2
Entity disambiguation (via description)
gpt-5-mini-2025-08-07
Target entity: Tacotron Target entity description: Tacotron is a neural network-based text-to-speech system that generates natural-sounding speech by predicting mel-spectrograms from text, often used in conjunction with neural vocoders like Parallel WaveNet.
-
A.
WaveRNN
WaveRNN is a neural network-based audio waveform generator designed as a more efficient, real-time alternative to earlier autoregressive models for tasks like text-to-speech synthesis.
-
B.
WaveNet
WaveNet is a deep generative neural network architecture for raw audio that produces highly natural-sounding speech and other audio signals.
-
C.
WaveGlow
WaveGlow is a flow-based generative neural network model for fast, high-quality text-to-speech audio synthesis.
-
D.
Parallel WaveNet
Parallel WaveNet is a neural vocoder architecture that accelerates high-fidelity audio waveform generation by distilling the autoregressive WaveNet model into a fast, parallelizable form.
-
E.
Wav2Vec2
Wav2Vec2 is a self-supervised deep learning model for automatic speech recognition that learns powerful audio representations directly from raw waveforms.
- F. None of above. chosen
Statements (49)
| Predicate | Object |
|---|---|
| instanceOf |
end-to-end TTS model
ⓘ
neural text-to-speech system ⓘ sequence-to-sequence model ⓘ |
| advantage | more natural prosody than traditional TTS ⓘ |
| architectureComponent |
attention module
ⓘ
decoder ⓘ post-processing network ⓘ text encoder ⓘ |
| basedOn |
attention mechanism
ⓘ
neural networks ⓘ sequence-to-sequence learning ⓘ |
| canBeUsedWith |
Parallel WaveNet
ⓘ
WaveNet ⓘ WaveRNN ⓘ neural vocoders ⓘ other neural vocoders ⓘ |
| comparedTo |
concatenative TTS
ⓘ
parametric TTS ⓘ |
| developedBy |
Google
ⓘ
Google Brain ⓘ |
| field |
deep learning
ⓘ
natural language processing ⓘ speech synthesis ⓘ |
| goal | generate natural-sounding speech ⓘ |
| improvedBy |
Tacotron 2
ⓘ
linked to:
Tacotron
|
| influenced |
FastSpeech
ⓘ
Glow-TTS ⓘ Tacotron 2 ⓘ Transformer TTS models ⓘ |
| input | text ⓘ |
| language | English ⓘ |
| output |
mel-spectrograms
ⓘ
speech features ⓘ |
| paperTitle |
Tacotron: Towards End-to-End Speech Synthesis
ⓘ
linked to:
Tacotron
|
| publicationType | research paper ⓘ |
| publishedBy | Google researchers ⓘ |
| replaces | traditional TTS pipelines ⓘ |
| requires | GPU acceleration for training ⓘ |
| supports | end-to-end training from text to spectrograms ⓘ |
| task | text-to-speech synthesis ⓘ |
| trainingData | paired text-speech data ⓘ |
| trainingObjective | minimize reconstruction loss on mel-spectrograms ⓘ |
| uses |
CBHG module
ⓘ
Griffin-Lim vocoder ⓘ attention-based decoder ⓘ character-level input ⓘ convolutional layers ⓘ encoder-decoder architecture ⓘ recurrent neural networks ⓘ |
How these facts were elicited
The pipeline generated the facts above by prompting gpt-5.1 with this entity's name + description and the instruction below.
Instruction
You are a knowledge base construction expert. Given a subject entity and a description of it, return factual statements that you know for the subject as a JSON list of dictionaries(triples), where keys must be "subject", "predicate" and "object". The number of facts may be very high, between 25 to 50 or more, for very popular subjects. For less popular subjects, the number of facts can be very low, like 5 or 10. # Requirements - If you don't know the subject at all, return an empty list. - If the subject is not a named entity, return an empty list. - Include at least one triple where predicate is "instanceOf". - Do not get too wordy. - Separate several objects into multiple triples with one object.
Input
Subject: Tacotron Description of subject: Tacotron is a neural network-based text-to-speech system that generates natural-sounding speech by predicting mel-spectrograms from text, often used in conjunction with neural vocoders like Parallel WaveNet.
Referenced by (3)
Full triples — surface form annotated when it differs from this entity's canonical label.
linked to: Tacotron
linked to: Tacotron