Japanese TTS Software Comparison. Good for Listening Practice? Can it do Pitch Accents?

Once you start immersion in Japanese after finishing beginner and early intermediate Japanese, you might wonder how you get that listening practice. How do I practice pitch accents? Sure, people will say you shouldn’t use TTS since it sounds unnatural. But Japanese TTS has improved drastically in recent years.

The cloud service TTS can be a hit or miss with its handling of TTS. However, TTS engines focus primarily on Japanese, namely AI Voice (the successor to Voiceroid), Cevio AI, and VoicePeak handles Japanese better. From my experience playing around with them, they handle pitch accents accurately off the fly. Also, buying a voice bank is more attractive than paying a cloud provider, especially if you plan to create a lot of audio for listening practice since it’s a one-time purchase. I want to share this since many of these Japanese voice synthesizers are on sale, and it might be worth looking at if you are learning Japanese.

From experience, VoicePeak sounds the most natural since it takes its engine from Synthesizer V, which is generally used to create covers and original songs. Also, it’s cross-platform too. AI Voice and Cevio AI requires Windows, and not running in a virtual machine, especially with the former.

What is a pitch accent? Depending on the word, it will have a different pitch on how you say it. This is the only context one has when listening to someone speak Japanese compared to reading. While the word may have the same reading, it will usually have different pitch accents.

Note that if a pitch accent sounds off or incorrect, you can use the NHK Pitch Accent or Japanese-only dictionary to get the correct one and correct it manually.

Here is a comparison of all the pitch accents from AI Voice, Cevio AI, and VoicePeak.