I agree technology has made video translations online very accurate but all depend on language complexity and what kind of software you are using. Popular services such as Google Translate, responsible for translating video captions on YouTube, have accuracy rates of 85% at most in languages like English—Spanish, French or German. Conversely, languages that are less well-spoken or have complex syntax, like Japanese or Arabic, experience a decrease in accuracy in the order of 10–15%.
Human-aided translation services are more reliable for brands and creators. Services like Rev. com promises up to 99% accuracy using AI translation, reviewed by human translators—a capability essential for industries demanding strict language requirements such as healthcare and law. Human-aided translation studies have pointed out that a translated work runs the risk of being misunderstood if it is not overseen by human validators. Netflix translates its content into over 20 languages, and relies on machine translation coupled with human editing to maintain cultural nuances that are essential for global audiences.
With real time translation, tools like Microsoft Azure Cognitive Services claim rapid processing speeds -- currrently less than 5 seconds per sentence for all languages with Latin script and at a fidelity level of (approx) 90%. On the other hand, being a real-time tool might also mean sacrificing accuracy for speed and it reaches to the viewer poorly. Nevertheless, anything with an emphasis on interactivity and several e-learning and gaming applications are still attracted to real-time.
Machine translation only services are much less expensive, as the above mentioned provider by itself forums around $0.10 per minute of video. In comparison, a human-aided service will typically range from 5–10 cents per minute all the way up to $1.50 per minute — a price point many companies would still consider acceptable given the increase in precision. In the case of exclusive, specialized content, industry experts like Dr. Anne Curzan — a linguistics professor at the University of Michigan and a member of their Partnership on Technology-Enhanced Language Learning and Instruction still argue now that “accuracy in translation is not just about words; it's about meaning and intent”, so human quality control remains very valuable.
Are online video translations reliable? Quality requested, language pairing and platform are the main factors here that basically determine how accurate something may be. Companies are still making choices about what tools to use, depending on which audience and budget they need for their trading strategies against cost versus precision trade-off. As a partial solution, hybrid tools combining the best of machine precision and human insight have been developed to produce good results for multiple languages leading users interested in balanced options to look at translate video online.