Audio recordings of the final conversations between pilots of the missing Malaysian jet and teams of air traffic controllers on the ground were "edited" before they were made public, voice experts say.
The tapes also appear to be recorded by at least two different audio sources, one of which may have been a digital recorder held up to a speaker, they said.
The analysts cautioned that their observations don't necessarily imply anything about the investigation into the missing flight.
The quality and brevity of the interactions between the cockpit and controllers made it impossible to glean any information about the pilots' state of mind before the plane disappeared, or even to determine whether both the pilot and co-pilot were speaking or if just one can be heard.
The audio recordings were published Thursday for the first time as part of a preliminary report by Malaysian authorities. In the report, Malaysia's Air Accident Investigation Bureau said a lack of real-time tracking devices caused "significant difficulty" in the hunt for MH 370, which disappeared March 8.
Analysts who listened to the recordings for NBC News did not know why they were edited, but discovered at least four clear breaks in the audio that indicated edits.
"It's very strange," said audio-video forensic expert and registered investigator Ed Primeau of Primeau Forensics, who has analyzed hundreds of audio recordings. He said the beginning and end of the recording are high-quality with a low noise floor, meaning ambient background noise is almost silent, unlike the middle.
"At approximately 1:14 (a minute, 14 seconds into the audio, which can be heard here), the tone of the recording change to where to me, it sounds like someone is holding a digital recorder up to a speaker, so it's a microphone-to-speaker transfer of that information. That's a pretty big deal because it raises the first red flag about there possibly being some editing," he said.
The next part that raises questions is two minutes, six seconds in, through two minutes, nine seconds in, he said.
"I can hear noise in the room, along with the increase in the noise floor. I can hear a file door being closed, I can hear some papers being shuffled. so I'm further convinced that, beginning at 1:14 continuing through 2:06 to 2:15, it's a digital recorder being held up to a speaker."
Long gaps in the communication throughout the recording also imply some editing, he said.
"But yet, at 6:17, there's a huge edit because the conversation is cut off. It's interrupted. And the tone changes again," he said. "The noise floor, when you're authenticating a recording from a forensic perspective, is a very important part of the process. All of a sudden, we go back to the same quality and extremely low noise floor that we had at the beginning of the recording."
Kent Gibson, a forensic audio examiner with Forensic Audio in Los Angeles, added that there appear to be additional edits at 2:11 and 5:08, and agreed it sounded as though the middle section was recorded with a microphone near a speaker.
"You can hear, at 4:07, pages turning or a person breathing, which is unusual," he said.
While it's not uncommon for the background of a recording to change when a cockpit communication turns over from ground control to air controllers — which happened about four minutes into this recording — that doesn't explain the noises that are heard.
"It's not unusual that there would be clicks when they push the button on the microphone, but it's very unusual to have a disturbance. Normally you wouldn't have any background," Gibson said.
A cut-off word also isn't out of the realm of possibility, he said.
"It wouldn't be unthinkable to have a truncated word because if somebody let go of the trigger on the microphone, it might cut off their word," he said. "But it would be very unusual to find a background differential at the same time, suggesting that Malaysian authorities or whoever presented this made edits for whatever reason."
Gibson said it’s possible the tapes could have been edited by Malaysian authorities "if the pilot dropped a hint that they didn't want to get out, if he said something that doesn't fit with the Malaysian government's party line."
But, he said, "It's more likely to be an inadvertent thing. But it's not the way to handle evidence."
The recording also could have come from different sources, he added.
"You can assume that the recording while they're still on the ground came from the tower and then you could assume that the communication with air controllers was while they're in the air," he said. "They may have just mishandled the cobbling of it together."
This doesn't necessarily prove anything about the investigation, he added.
"Unfortunately, there are no smoking guns, except there are edits. And there are clear edits," he said.
Tom Owen, a consultant for Owen Forensic Services audio analysis and chairman emeritus of the American Board of Recorded Evidence, said edits were to be expected.
"There's things that have to do with timelines and radar that they have available, but they don't make them available," he said. "They wouldn't give you anything that would be enlightening for the public to any secretive information. I don't see that as a problematic issue."
"This is not a good maneuver or a good faith move by the Malaysian government because of all these questions with regard to the different anomalies and edits that are in this recording," he said.
Audio experts felt the quality of the transmissions was too low to offer analysis of the pilots' voices.
Forensic audio expert Paul Ginsberg said even after enhancing and slowing down the conversations, there wasn't good enough, or long enough, sound samples to make a determination on the pilots' stress level.
"It's analogous to blowing up a photograph. It's the same amount of information," he said. "I don't know that any such determination would be admissible."