Speech-to-speech translation is limited by language processing, not signal processing

Evidence

Speech recognition has already been adopted in many applications. For example, some televisions and handheld devices can already be controlled by voice, and telephone customer services may use speech recognition in automated tasks. Speech recognition is effective in browsing the contents of spoken speech databases (Chelba et al., 2008). One application that is also actively being studied is a telephone service that translates the language of a speaker into another language. Although this service has not yet been achieved, translation in limited cases has been possible. Optimally, the characteristics of the voice of the original speaker should be preserved in the translated speech output. The biggest challenges are, however, in the processing of language, and not in the processing of speech signals. — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 2275

Categorization

Category: Factual Claims

• XREF: Connects to the reader’s interest in acoustic communication applications — speech translation sits at the boundary of acoustics and linguistics; parallels computer translation barriers discussed in other tech literature.
• SEED: Could form the spine of an article on where acoustic-communication technology succeeds versus where it stalls.

Related

• The hardest part of speech-to-speech translation is language processing, not signal processing. (similarity 0.92)
• Beyond intelligibility, telephony must preserve speaker recognizability (similarity 0.81)
• Synthesis and recognition demand language knowledge databases that coding does not. (similarity 0.81)
• Language knowledge beyond acoustics is required for automatic speech recognition. (similarity 0.79)
• Human speech evolved as a robust channel tolerant to high background noise (similarity 0.79)
• Speech transmission quality hinges on undistorted information transfer (similarity 0.79)
• Rule-based speech synthesis survives only where computational power is scarce. (similarity 0.79)
• Speech coding is the most mature speech technology because it needs no databases. (similarity 0.78)
• Speech production can be modeled mathematically to derive useful signal processing algorithms (similarity 0.78)

Broader Connections

• Linguistic context lets listeners understand speech at lower acoustic intelligibility #corroborates (0.82) — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 2430