Towards speech-to-text translation without speech recognition
This addresses the problem of speech translation for low-resource languages where traditional methods are unavailable, but it is incremental as it builds on existing unsupervised techniques and shows limited performance.
The paper tackled speech-to-text translation in low-resource settings without ASR or MT by using unsupervised term discovery to create pseudotext from audio, paired with translations to train a bag-of-words MT model, achieving correct translation of some content words in test data despite low recall from cross-speaker challenges.
We explore the problem of translating speech to text in low-resource scenarios where neither automatic speech recognition (ASR) nor machine translation (MT) are available, but we have training data in the form of audio paired with text translations. We present the first system for this problem applied to a realistic multi-speaker dataset, the CALLHOME Spanish-English speech translation corpus. Our approach uses unsupervised term discovery (UTD) to cluster repeated patterns in the audio, creating a pseudotext, which we pair with translations to create a parallel text and train a simple bag-of-words MT model. We identify the challenges faced by the system, finding that the difficulty of cross-speaker UTD results in low recall, but that our system is still able to correctly translate some content words in test data.