Fresh Ears on The Inland Sea’s Exquisite Soundscapes

Inside Criterion / Tech Corner — Aug 23, 2019
Criterion audio supervisor Ryan Hullings, photographed by Grant Delin

When director Lucille Carra and producer/editor Brian Cotnoir set out to make The Inland Sea (1991)—a poetic travelogue just released last week in a brand-new Criterion edition—they had a strong vision for how they wanted their film to sound. They knew they wanted the viewer to hear the movie’s journey as they’d experienced it, as a tapestry of enveloping sounds: the low churn of a ferry’s engine, the volley of distant bird calls in a serene forest, the hectic din of human activity in a busy fish market. During the film’s three-and-a-half-week shoot, Carra, Cotnoir, and cinematographer Hiro Narita traveled around the islands of the Inland Sea—the body of water surrounded by the Japanese main islands of Honshu, Shikoku, and Kyushu—observing the rituals and textures of daily life in the remote region, a trip inspired by author and Japanese-cinema expert Donald Richie’s classic 1971 memoir of the same title. Also along for the journey was soundman Tom Hartig, who recorded the audio nonsynchronously and in stereo, so that the film’s soundscape would be as fluidly immersive as its arresting visuals. In postproduction, two additional aural elements would be added to the mix: a reflective score by Woman in the Dunes and Ran composer Toru Takemitsu and a scene-setting voice-over by Richie himself, reading passages he had adapted from his own book.

After it was in the can, The Inland Sea went on to enrapture viewers, garnering acclaim at Sundance and other festivals, and during a subsequent commercial run through venues that included the venerable New York art house Film Forum. But when it came time to begin preparing the film for its Criterion edition, audio supervisor Ryan Hullings immediately noticed that things were nonetheless not quite sounding as they should. After reviewing the film’s master, Hullings identified two issues in particular that went beyond the usual audio cleanup that Criterion performs in advance of its releases, which includes attenuating noise and analog distortion and manually removing impulse noises—like clicks, pops, and thumps—resulting from the physical deterioration of film elements. “The Richie VO had a number of problems with it,” says Hullings, “and the stereo audio that had been recorded in the field was very unbalanced and out of phase.”

Carra and Cotnoir had all along been aware of problems in the voice-over. Richie’s narration happened to have been recorded in two different settings with two vastly different recording setups—a more informal session in which a single microphone was used, and a more professional studio session in stereo, though with one of the two microphones of a noticeably inferior quality—leading to a narration without a uniform sound. In addition, the voice-over had been edited in such a way that the track cut out altogether when Richie was finished speaking, where typically the “room tone”—the particular quality of the silence of the room where the recording had taken place—would carry on, on its own, in the background. After The Inland Sea showed at the Hawaii International Film Festival, where it was awarded the best documentary prize, Roger Ebert publicly expressed his love for the film, but privately took Carra aside to talk about its soundtrack. “He said, ‘Well, you know, there’s a sound problem,’ and it was the audio dropout on Donald’s narration,” Carra says. The film went into its theatrical release—and later on its release on VHS, laserdisc, and then DVD—with this problem unresolved.

Combing through the soundtrack’s dialogue “stem”—the isolated audio track that in this case contained the voice-over—in Avid’s Pro Tools program, Hullings set about smoothing over these distracting qualities, a painstaking process that wound up taking him many hours. After he separated out the better-sounding microphone from the stereo session, so that the voice-over would be monaural across the board (though it would ultimately be presented in dual-mono fashion, across both channels), it still “tonally sounded nothing like the mono recording session,” he says. At that point, a little “EQ-matching”—or digitally equalizing the frequencies of the two separate sessions—did the trick. Hullings’s final step was editing in some additional room tone, eliminating the dropouts that Ebert had flagged back in the early nineties. “The voice-over doesn’t throw you off now,” Hullings says. “It sounds like the same person talking the whole time, in the same place.”

The video clip below demonstrates the subtle but nonetheless palpable difference. The first half presents a moment from the film’s original soundtrack that draws from both the informal and the professional voice-over recording sessions during what is intended to be a continuous passage, while the second half features the same moment with the restored audio. (Headphones are recommended for both illustrative videos in this article.)

A portion of the film’s voice-over before and after restoration

Finding a solution to the stereo issue, however, proved to be much more vexing. When Hullings initially listened to the audio that the filmmakers had recorded in the field, “it sounded as if the left channel was totally unrelated to the right channel a lot of the time—there would be waves in one ear and wind sounds in the other, whereas if you’re recording with a normal stereo setup, you’d get an image where the left and right were at least related to each other, in some sort of balance,” he says. “This didn’t create a stereo field in any way; it’s basically these two very foreign sounds.”

After testing out a few potential fixes, including some sophisticated phase manipulation, and failing to get anywhere, Hullings decided to try something on a lark. “I wondered if the stereo had been recorded in this somewhat esoteric fashion that I had been experimenting with for music: mid-side recording,” he says. “There’s a lot of music production that uses it, and it’s neat, but I’d never encountered it in a final mix for a film.” Instead of the typical “XY” stereo setup, in which two directional microphones are aimed toward the sound source and angled at 90 degrees to each other, the mid-side technique involves placing a “mid” microphone directly facing the source and a bidirectional “side” microphone perpendicularly below it—an alignment that ultimately captures richer sonic information that can then be manipulated in the sound design, unlike the results of XY recording. In order to achieve a suitable stereo image using the mid-side method, though, the tracks have to first be “decoded.” And at an impasse as to how to bring the film’s sound into proper balance, that’s just what Hullings tried doing. “I tried decoding the movie’s field-recorded audio as mid-side rather than treating it like a typical stereo track, and it immediately clicked,” says Hullings.

In the below video, you can hear the pre- and post-restoration audio for an early scene that depicts a visit to a hilltop Shinto shrine on a particularly peaceful Inland Sea island. The first half of the video demonstrates the disconnect in the original mix: the three consecutive sounds that punctuate the prayer of one visitor to the shrine—the coin drop, bell ringing, and hand clapping—are mostly present in the left channel, while ambient forest sounds dominate the right channel. The second portion shows the prayer audio in its properly decoded form, the coin drop, bell ringing, and hand clapping now better balanced in the right and left channels, creating a more accurate stereo image—and thus bringing the viewer into an ever-so-slightly closer relationship with the environment on-screen.

A portion of the field-recorded stereo audio before and after restoration

With Hullings’s work on decoding the field-recorded audio, the stereo image all of sudden came into sharper focus. It turns out that, nearly three decades ago, Hartig (now a regular member of Quentin Tarantino’s sound team) had taken the extra step of recording sound for the film using the mid-side method, which would allow the filmmakers more leeway to tinker with the width of the stereo field in postproduction. But due to miscommunication, and a postproduction period during which Carra and Cotnoir were focused on simply getting their low-budget labor of love to the finish line in a timely fashion, this audio had never been properly mixed. And only now, with the Criterion release of The Inland Sea, would the stereo results be heard as they were always intended to be heard.

With the hushed beauty of its visuals now married to a richer, more fully engrossing stereo soundtrack and a more cohesive-sounding voice-over, then, The Inland Sea in its Criterion restoration is a more transporting film than ever before—and Carra and Cotnoir couldn’t be happier about it. “Brian and I came in, and we saw some sections of what Ryan was working on, because he wouldn’t do anything without our okay, and I said, ‘Oh, this is beautiful! It adds a whole new dimension to the film,” says Carra. For his part, Cotnoir—who had sought, as he was editing the film, to accentuate the dreamlike quality that the nonsynchronous field-recorded sound created in conjunction with the images—finally had the feeling of “artistic intent realized” when he and Carra had the opportunity to spot-check the audio work Hullings had done with their approval. “It was a great film then, and it’s an even better film now,” Cotnoir says. “I can’t wait to see this whole thing projected on a big screen in stereo. I kind of want to have a party when this thing comes out.”