Last week I attended a very interesting webinar from one of our vendors (JBI Studios*), about machine-generated voices for voiceover projects. In general, this new technology is recommended for e-learning courses and short announcements such as airport announcements or telephone customer service. Here at LCW, we specialize in the translation and publishing of e-learning courses, and for us, coordinating the voiceover talent and/or inserting subtitles is always one of the most time-consuming parts of these projects, beyond the translation itself.
The webinar was very informative on the pros and cons of using this technology; it made me think about how and when this might be used on our e-learning projects. The webinar also raised awareness about what to look for while working on these projects. Here is what I found:
- Lower minimum session costs, and post-production costs since there are fewer people and resources involved. Also, there is a lower turnaround time due to shorter “recording” times (compared to human voice talent).
- At first glance, it may seem that this technology is cheaper due to the fact it requires very few human resources and interactions. However, our vendor recommended making sure we do a cost analysis prior to considering machine-generated voices, because sometimes the cost may be higher than using “human” voices (due to initially higher time investment to set up the technical aspects of the project).
- Besides having to create a phonetic list, one of the setbacks includes having to “train” these voices for tricky pronunciations. If the training is rather extensive, this may not be a cost effective option.
- Costs are coming down slowly but surely.
- The faster turnarounds are a result of: the absence of session scheduling, no recording breaks, faster postproduction etc…
- Turnaround time is probably one of the main reasons to choose machine over human voices.
- Some clients had complaints about the voices sounding ‘unnatural’, ‘non-human’, and even ‘irritating’. Computer-generated voice can eventually sound awkward when listened to for an extended period of time.
- Currently, there is a limited language set available on the market including most European languages (Russian included), Japanese, Korean, Chinese Mandarin, and Arabic. Though this list continues to grow every day.
- Like we mentioned before, using computer-generated voiceovers may not be your cheapest option.
- Lastly, make sure your vendors have the right licenses; the machine-generated voices you often here are usually proprietary. There are various options for free voiceover talent available online, BUT these options are usually not available for commercial use.
In conclusion, make sure you consider all these variables before deciding to replace your voiceover talent with machine-generated voices. This technology will be the most cost-effective option for smaller projects with low word-counts, multiple languages, and multiple voices. But, keep in mind that technology is constantly improving so these recommendations may change. It’s not likely that machine-generated voices will ever completely replace human voice talent, but eventually a hybrid approach is likely.
*JBI Studios is a full-service audio, video and multimedia production house specializing in foreign languages.