The documentary broadcast by Netflix contains an audio recording by an ex-CGM member, Maple Yip, who claimed that she had secretly recorded the audio while being raped by President Jung. Based on the findings of an independent forensic laboratory, there is strong evidence to show that Maple Yip, who provided the audio recording, had lied about the source of the recording. Further analysis also shows that the recording itself had been edited and is not authentic.
The recording purports to show President Jung asking Maple if she had cum, and President Jung saying that he had cummed 50 times. It had been submitted as proof in the Korean court and also used as the key evidence in the Netflix series.
In order to ascertain the authenticity of the audio recording, Taiwan CGM engaged an authoritative voice-related research institute [(U.S. business-related) MuScene Voice Forensic Laboratory] to conduct forensic testing on the recording. The (U.S. business-related) MuScene Voice Forensic Laboratory is an impartial third-party laboratory commissioned by law enforcement agencies for identification, investigation, and laboratory testing. It conducts inspections commissioned by the courts, prosecution agencies, and national security teams (the military) from all around Taiwan and the US (including CIA, FBI, State Department, Secret Service). It has a rich academic base, a large amount of forensics hands-on experience, and cutting-edge technical knowledge. It is an authoritative research institute that conducts in-depth research on sound synthesis and modulation testing and identification, sound signal enhancement, acoustic fingerprint-sound identification, file conversion and sound compression through scientific research, and data analysis methods.
After detailed analysis using waveform data analytics, the results revealed that the voice recording was edited in a disjointed fashion. It also revealed that the audio was not downloaded from iCloud, contrary to Maple’s claim that she had lost the original file but retrieved the backed up copy from iCloud (where iPhone-generated media data are backed up). The forensic report showed that the voice recording used in the Netflix film did not have formatting or characteristics of iPhone origins and was not downloaded from iCloud. Audio files that are recorded using an iPhone app or the default audio recorder would contain inherent characteristics, but the Netflix voice recording does not have any of those characteristics.
Here is the detailed forensic analysis:
On March 3, 2023 Netflix released the documentary In the Name of God: A Holy Betrayal. In the 3rd episode of the series, titled JMS, Messiah with electronic anklets, from the 40:00 to 41:07 mark, an audio file claimed by Maple Yip to be the recording of an alleged sexual assault (henceforth called “Netflix Maple audio file”) was examined for authenticity by a third-party Voice Forensic Laboratory, and compared with the audio file previously aired by Korean media JTBC News on July 12, 2022 Jung Myung Seok Voice Files (정명석 ‘육성파일’) (henceforth called “JTBC News audio file”). The following are the findings:
Evidence that Netflix Maple lied about the source of the audio file that was produced and broadcast on the Netflix platform
(Reference URL: https://reurl.cc/GeG49y)
- An environmental sound spectrum evaluation revealed a dark band at 8 kHz which would not be generated in the room mode* of a spatial environment. The spectrogram above 8 kHz was also very clean with no traces of spatial noise found in normal spatial recordings. Hence, it is inferred that the original frequency response of the Netflix Maple audio file should be 8 kHz and the sample rate should be 16 kHz. (Figure 1)
Figure 1: The sound spectrum of Netflix Maple audio file.
- During the South Korean court trial on February 13, 2023, Maple claimed that the audio file was uploaded to iCloud for backup before the phone was sold. It is reasonable to infer that Maple had used an iPhone to record the audio. However, according to the spectrum data verified by the Voice Forensic Laboratory, it can be seen that the frequency response of the Netflix Maple audio file is not consistent with 16 kHz and that this evidence (Figure 2) is also inconsistent with the iPhone native app recording spectrogram characteristics (Figure 3). It can be confirmed that the recorded file of the Netflix Maple audio file was not recorded by an iPhone native app.
Figure 2. Spectrum diagram of Netflix Maple audio file verified by the Voice Forensic Laboratory
Figure 3. Spectrum diagram of the iPhone native app, which is characterized by a frequency response that can only reach 16 kHz
- There are two main conversion methods for digital audio files: resample** or convert. The conversion method used by Netflix’s professional audio-visual platform is mostly “resample”. Aside from Netflix’s conversion of the Netflix Maple audio file using their professional audio-visual platform to resample, it is assessed that before being provided to the Netflix platform, the file had gone through “convert”. The sampling rate was likely to be 16 kHz, then converted to 44.1 kHz, finally resampled to a 48 kHz audio file by Netflix’s post-production. The sound spectrum characteristics of a typical “resampled” file using professional audio-visual post-production software is shown in Figure 4, while the sound spectrum characteristics of a “converted” file is shown in Figure 5. The sound spectrum characteristics of Netflix Maple audio file is shown in Figure 6, which is similar to Figure 5, but not similar to Figure 4. This evidence shows that the Netflix Maple audio file had gone through “convert” processing, and it cannot be ruled out that the file could have been modified or edited. Hence, it is doubtful that one could show or restore the content of the original recording of the Netflix Maple audio file.
Figure 4: The special sound spectrum characteristics of “resample”
Figure 5. The sound spectrum of “convert” will generate mirrored sound information
Figure 6: Spectrum diagram of the Netflix Maple audio file [Footnote 1]
Comparing the recording content of the Netflix Maple audio file with JTBC News, the recording content of the second audio file raises suspicion that the Netflix Maple audio file had been altered.
- The JTBC News audio file was originally “God chose you like this; it is God’s will.” In the Netflix Maple audio file, the last sentence was deleted, and the content was changed to “God chose you like this, so don’t touch other people’s… ” at the 40:59 mark of the Netflix series (Video timing of 0:11‒0:21 in Footnote 2’s link)
- The JTBC News audio file was originally “Don’t touch other people’s “this”, because it belongs to God. Do you understand? God is the teacher’s, and it is God’s will. Do you understand? This is God’s will.” Then, the Netflix Maple audio file was changed to “Don’t let others touch this (woman’s moaning sound added) because this is God’s (woman’s moaning sound added).” (Video time 0:30‒1:39 in Footnote 3’s link)
- The JTBC News audio file was originally “Don’t touch other people’s this, because it is from God. Do you understand?” A male responded with “네” (yes), but the Netflix Maple audio file deleted the male’s response and added a woman’s moaning” (Footnote 2 video time 2:09‒2:17, note 3 video time 1:40‒2:15)
- At 40:31 and 40:32 in the Netflix Maple audio file, in addition to the sound of “쌌어?” (Have you had diarrhea/cummed?)”, there were two consecutive male exhalation sounds at the same time. In the same span of time and space, a person cannot simultaneously exhale while speaking. So it is reasonable to infer that there was a third person at the audio scene of the alleged sexual assault. (Video time 2:16‒2:56 in Footnote 3)
The above mentioned frequency spectrum analysis showed that Maple had lied about the source of the audio recording, which was not from an iPhone. This is of particular significance because she had attempted to admit this recording as the only physical evidence of the alleged crime, but the Korean legal forensic team were unable to verify the authenticity of the recording, as the original iPhone that Maple allegedly recorded the audio was no longer available. The detailed analysis clearly showed that the recording was never recorded on an iPhone and that it had undergone multiple stages of resampling and conversion.
Further analysis of the audio waveform itself clearly showed that this audio file is not authentic and has been edited. It cannot be used as evidence, let alone be used to guide public understanding of what actually happened.
* A room mode is essentially a “bump” in a room’s frequency response that is facilitated by the room’s dimensions and the way those dimensions cause soundwaves to interact with each other. There are three types of room modes: axial, tangential and oblique.
**Resampling, in terms of audio files, is also known as Sample Rate Conversion. This is done when you need to convert a digital audio file from a given sample rate into a different sample rate. Sample rate is the number of samples of audio carried per second, measured in Hz or kHz (one kHz being 1,000 Hz).
[Footnote 1] This recording file has been verified by forensic laboratories. The dark band with a frequency of 8 kHz in the sound spectrum is not present in the room mode* of a spatial environment.
[Footnote 2] JTBC News URL: https://www.youtube.com/watch?v=LstcHXUHZy0
[Footnote 3] Altered comparison video URL: https://youtu.be/pHKwXspdVok