Last updated October 2, 2023
This page serves as a compilation of corrections for any factual inaccuracies or errors across all video uploads on the Junferno YouTube channel. This includes errors in both the video transcript and visuals, but does not account for things like graphical glitches or typos in the captioning. These corrections are also included in their respective video's descriptions. Corrections on this page will be sorted by video in reverse chronological order of the video's publishing date.
To report a correction please reach out by email at email@example.com.
List of Corrections
- [Clarification*] A common misconception is that Turing claimed that no machine could compute something that a Turing Machine can not. Turing never explicitly stated this, as he only proposed a definition of "effective computability", but it is a topic still debated on today. While there are theories for quantum "hypercomputation", quantum computing as we know it (Deutsch 1985 presented at 13:34) has the same computing power as Turing Machines, just faster. Deutsch, however, does claim that his Church-Turing-Deutsch principle encompasses every physical process.
- [Visual error at 6:02] "Ultrecht" should be Utrecht
- [Visual error at 10:32] Shortest path from A to C should be ADC, yielding a distance of 1+7=8, making the total distance from A to C to E: 8+12=20
- [Visual error at 25:09] "Nvidea" should be Nvidia
- [Footnotes] Footnote previously stated that OpenSeeFace's models are trained by transfer learning from MobileNet V3. This is incorrect. OpenSeeFace's models are trained from scratch using MobileNet V3 architecture as a backbone. MNV3 is a ConvNet designed for mobile phone CPUs. The final landmark determination step also doesn't use normal linear regression, but a heatmap based approach with offset vectors, where UNet upscaling layers follow after the MNV3 feature extraction. Thank you @Emiliana_vt for this detail.
- [4:52] Timbre should be pronounced like "tamber" (tăm′bər) instead of "timber".
- [12:33] Due to contradictory references to the Daisy Bell recording in whether the synthesis was done on the IBM 7094, 704, or 7090, it is possible[author's conjecture] that the synthesis was demonstrated on multiple computers throughout its development. However, there is concrete evidence that it was at least demonstrated on the 7090, as stated by developers John Larry Kelly Jr. and Carol C. Lochbaum at a 1962 congress.
- [12:33] Although this segment is about the source-filter model, the model that synthesised Daisy Bell was actually either an "artificial vocal tract" (as described by Kelly and Lochbaum in 1962. Kenmochi Hideki called it an "acoustic tube model" in 2014) or a "terminal analog" (Kelly and Lochbaum in 1962). Both comprised of a large database of parameters that applied ad-hoc rules given a sequence of phonemes, pitches, and timings, with the first model using articulatory parameters (e.g. nasal, labial) while the second used acoustic parameters. While both it and the Voder were products of Bell Labs, the two had limited further relation.
- [2:27] Modern Polish typists use the programmer's keyboard as opposed to the standard one. A better example of a keyboard that uses separate keys for special characters is the Swedish keyboard.
- [2:58] In the French AZERTY, the grave-accented a (à) has its own key (though the US International layout uses a dead key). A better example of a letter using a dead key on the AZERTY layout would be the circumflex-accented a (â) which is typed by pressing the '^' key followed by the 'a' key.
- [13:44] On the Korean 3-set keyboard, the initial consonants are on the right and the final consonants are on the left.
- [14:24] The Romaji for 今日は is usually "kyouha" in modern Japanese, meaning "today". "Konnichiwa" (or "konnichiha") is written with the Hiragana characters こんにちは.
- [14:49] JIS stands for Japanese Industrial Standard, not Japanese International Standard.
- Keyboard Layout Info. (n.d.). Kbdlayout.info. Retrieved December 2, 2021, from http://kbdlayout.info/
- JISC-Japanese Industrial Standards Committee. (n.d.). Www.jisc.go.jp. https://www.jisc.go.jp/eng/
- Kelly, John Larry Jr. and Lochbaum, Carol C. (1962) Speech Synthesis. In Proceedings of the Stockholm Speech Communication Seminar., pages 1-4, Copenhagen, Denmark. Paper G42
- Kenmochi, Hideki and Fujimoto, Ken (2014). ボーカロイド技術論: 歌声合成の基礎とその仕組み [Vocaloid Technological Theory - Foundation and mechanism of singing voice synthesis]. Tokyo: Yamaha Music Media.