Differentiation through limited recorded voice data, a diverse range of speakers, Custom Voice, and end-to-end delivery
First written in FY2018. 3 of 7 entries say "Advanced", first in FY2021. Latest entry: 2026-06-25.
No other company in this theme has enough history to compare yet.
- 2026-06-25AdvancedOther Disclosures
「As a result, AITalk micro was adopted because it met the requirements for embedded devices, including lightweight performance, clear voice quality and flexible support for successor models as well.」
The adoption of AITalk micro in Panasonic AI-enabled video doorphones concretely demonstrated its differentiation for embedded devices: lightweight, low-resource operation, highly intelligible voice quality and flexible support for successor models. - FY2024RestatedAnnual report
「As of the filing date, we provide 18 speakers in total: 8 female and 4 male standard Japanese speakers, 2 boys, 2 girls, 1 female and 1 male Kansai dialect-style speaker.」
Continuing to disclose the existing differentiators of limited recorded voice data, 18 speakers, more than 400 Custom Voices and end-to-end delivery. - FY2023AdvancedAnnual report
「As of the filing date, we provide 18 speakers in total: 8 female and 4 male standard Japanese speakers, 2 boys, 2 girls, 1 female and 1 male Kansai dialect-style speaker.」
The number of available speakers increased from 16 previously to 18. The track record of Custom Voice creation also expanded to more than 400. - FY2022SoftenedAnnual report
「As of the filing date, we provide 16 speakers in total: 5 female and 5 male standard Japanese speakers, 2 boys, 2 girls, 1 female and 1 male Kansai dialect-style speaker.」
This filing lists 16 available speakers, a nominal decrease from the 17 stated previously. Custom Voice has meanwhile been maintained at more than 370. - FY2021AdvancedAnnual report
「As of the filing date, we provide 17 speakers in total: 8 female speakers, 5 male speakers, 2 boys and 2 girls.」
The number of available speakers increased from the previous 15 to 17, while the number of Custom Voices also expanded to more than 370. - FY2019RestatedAnnual report
「While a typical recording requires several dozen hours, equivalent to several thousand to 10,000 recorded sentences, we have achieved the creation of voice dictionaries from approximately 2 to 6 hours of recordings, equivalent to approximately 200 to 600 sentences.」
Continuing to identify limited recorded voice data, 15 speakers, more than 300 Custom Voices and end-to-end delivery as differentiating factors. - FY2018First statedAnnual report
「We are conducting research and development aimed at producing high-quality speech synthesis from limited recorded voice data. While a typical recording requires several dozen hours, equivalent to several thousand to 10,000 recorded sentences, we have achieved the creation of voice dictionaries from approximately 2 to 6 hours of recordings, equivalent to approximately 200 to 600 sentences.」
The company also identifies 15 speakers, more than 300 Custom Voices and in-house end-to-end support as strengths.