bangladesh2024justicewatch.org

Trial of Sheikh Hasina and others

Court 1

Case on 2/2025

Trial Day 14/21

9th/23rd Spet 2025

Back to Trial page

Court 1          Case no 2/2025              Trial Day 14/21          9th/23rd Sept 2025               Back to Trial page

Witness 45: Shahad Jubayer Lawrence

Evidence of Shahad Jubayer Lawrence

My name is Shahad Jubayer Lawrence, my approximate age is 30 years. My address is Village XXX Police Station XXX, District- Tangail.

Currently, I am working as an SI (Sub-Inspector) in the Digital Forensic Lab of CID Dhaka. On 12/09/2025 AD, a CD and a DVD were sent as evidence in this case by the Special Investigating Officer of this case. The CD contained the disputed conversation between former Prime Minister Sheikh Hasina and the then Vice-Chancellor of Dhaka University, Maksudur Kamal. The video contained the sample voices of both of them. Upon examination, I separated the female voice and the male voice from the conversation stored on the CD using tools utilized in the forensic lab. I conducted a comparative verification of the female voice and the male voice stored on the CD with the sample voices of former Prime Minister Sheikh Hasina and the then Vice-Chancellor of Dhaka University, Maksudur Kamal, stored in the video. Upon verification, a match was found between the female voice and the sample voice of former Prime Minister Sheikh Hasina, and a match was found between the male voice and the sample voice of the then Vice-Chancellor of the University, Maksudur Kamal. In addition, the Special Investigating Officer generally came and took away the transcription of the said conversation from our office. This is that copy of the opinion (Exhibit- 22). This is my signature on the copy of the opinion (Exhibit- 22/1). The main copy of the opinion is with me.

This is my deposition.

Cross-examination

The voice of one person might match that of another person.

It is not true that the statement that a match was found between the verified female voice and the sample voice of former Prime Minister Sheikh Hasina is false.

It is not true that the voice examined is not that of former Prime Minister Sheikh Hasina.

It is not true that the statement that a match was found between the male voice and the sample voice of the then Vice-Chancellor of Dhaka University, Maksudur Kamal, is false.

It is not true that I gave false testimony due to political and administrative influence.

Re-examination:

Mizanur Rahman: What standard tool do you use?

In case of voice examination, the international standard software LR Ratio (Score) used in the CID Forensic Lab is generally considered to be one and the same if the disputed voice and the sample voice are above 1. In this case, the LR Ratio (Score) of the sample voice and the disputed voice of former Prime Minister Sheikh Hasina was 2.7514. For this reason, I gave the opinion that the disputed voice is Sheikh Hasina’s. The LR Ratio (Score) of the sample voice and the disputed voice of former VC Maksudur Kamal was 4.64153. For this reason, I gave the opinion that the disputed voice is that of the former VC of Dhaka University, Maksudur Kamal. (With objection)

This is my deposition.

Re-Cross-examination

The result of my examination has been examined by no other higher authority. I do not recall the name of the higher authority that verified my result. The signature of the higher authority that verified my result is not present on the submitted report.

It is not true that I submitted a false report.

It is not true that I gave false testimony due to political and administrative influence.

23 September

Re-examination

After 3 videos were received by the Investigating Officer of this case at the Digital Forensic Lab, CID Dhaka on 1/06/2025, I began the waiting (forensic examination) process. One video contained 3 audio files. One of the 3 audio files had no conversation. From the conversation in the remaining 2 files, I separated the male voice and the female voice through Segmentation. According to the result of the forensic software, a match was found between the 2 male voices and Inu’s sample voice, and a match was found between the 2 female voices and Sheikh Hasina’s sample voice, and a match was found for the male voice.

According to the forensic software result, the LR Scores for the 2 male voices with the sample voice of JASHOD President Hasanul Haq Inu were found to be 19.1274 and 1.18064. According to the forensic software result, the LR Scores for the 2 female voices with the sample voice of former Prime Minister Sheikh Hasina were found to be 1025.9954 and 432.88127. Based on the information obtained during the examination, I provided the opinion that a match was found between the 2 male voices and the sample voice of JASHOD President Hasanul Haq Inu, and a match was found between the 2 female voices and the sample voice of the former Prime Minister. In the international standard audio forensic software used in the Digital Forensic Lab, if the LR Score is 1 or above, the concerned voices are considered to be the voice of the same person.

I provided the opinion along with the transcription of the conversation and handed over the DVDs to the Investigating Officer. This is that opinion provided by me. (Exhibit- 23). The opinion is described on pages 29-43 of the supplementary seizure list and documentary evidence. These are my signatures on the opinion (Exhibit- 23/1 Series).

This is my deposition.

Cross-examination

It is not true that I falsely said that a match was found between the 2 male voices obtained through segmentation and a match was found between the 2 female voices and the sample voice of former Prime Minister Sheikh Hasina.

I obtained BSc and MSc degrees in EEE from Dhaka University. After joining CID, I received training on forensic subjects.

It is not true that I did not properly examine the said voices.

It is not true that the audios sent to me were created using AI technology.

Yes, voice can be generated by AI technology. However, the authenticity of whether a conversation is AI generated can be verified by examination in a forensic laboratory. An AI-generated voice will never perfectly match a human voice.

It is not true that the opinion I provided is incorrect.

It is not true that I submitted this report under the influence of politics and administration.

Leave a Reply

Your email address will not be published. Required fields are marked *