a site by John Cobley

a coppice gate

Marni Nixon: The Singing Voice of Deborah Kerr, Natalie Wood and Audrey Hepburn

by John Cobley

Thursday Apr 28th, 2016



After re-watching the 1964 film My Fair Lady on an old videocassette, I did some research on the dubbing of Audrey Hepburn’s singing voice  by Marni Nixon.  I wanted to find out how she went about this delicate process. I learned that by the time she did My Fair Lady, Marni was very experienced, having worked on two major musicals, dubbing for Deborah Kerr in The King and I (1956) and for Natalie Wood in West Side Story (1961). How was this dubbing process done? What were the technical issues back in the 1950s? How co-operative were Hepburn, Kerr and Wood? How was Marni able to sing so convincingly in the voices of three very different characters? To find out, I ordered Marni Nixon’s book I Could Have Sung All Night from my local library.


The King and I

Marni Nixon was 25 when she was hired to dub for her first major film. She had been moderately successful in making a living as a singer in New York but had been unable to break into the top rank. So she had moved to LA. With some experience dubbing for Margaret O’Brien, Joan Crain, Lili Palmer, Ida Lupino, she felt confident enough to accept the job of dubbing all the six songs that Deborah Kerr’s character had to sing in The King and I. After this musical had enjoyed great stage success on Broadway, Twentieth Century Fox decided to purchase the film rights. It was to be a major production with a big budget.


Dubbing on such a large scale was unusual in Hollywood, so much of Marni’s preparation was original. “We were improvising,” she wrote. Her first step was to listen to a recording of Deborah Kerr’s speaking and singing voices—“so that I might assess her voice and get under her skin.” She found Kerr’s voice “breathy, straight and pure, almost like a boy soprano,” and felt “an immediate kinship” with it. Next, Marni had to record her own singing voice so that it could be approved by the show’s composer Richard Rodgers.



Her contract was for just six weeks, one week for each of Kerr’s songs. On meeting Kerr, she discovered that both their families had their origins in Scotland on the Isle of MacIntyre. The two women bonded immediately and worked co-operatively throughout the six weeks. They worked independently on each week’s song in the morning, learning it first and then trying to perfect it. In the afternoon they sang together in a mock set that duplicated the scene that would eventually be filmed. First there was some rehearsing with a pianist, where they would begin to see “how the other breathed and phrased the lyric.” Then they would rehearse the song physically on the mock set. Kerr would move as she was choreographed to do in the upcoming filming, and then when she began singing Marni would “follow her around like a shadow,” copying her movements and stance. “After a while we started to look like twins,” Marni has written. “I would imagine myself in her body.”


The penultimate step in each week’s work was to record the song with a 50-piece orchestra. Both women were placed inside a sound-proof room with a window connecting them visually with the conductor. Each recorded the song separately. For Marni’s recording, a new gadget was used, a voice modifier that could “emphasize the lower partials” of her voice. Several times both their voices would be used in the same song, so their voices had to be indistinguishable. For example, in “Shall We Dance” Kerr “talk-sang” some of the number, while Marni’s voice was used when “the singing got tougher and more sustained.” 


Last, before the final sound mixing, Kerr would attempt to lip-sync to the playback of Marni’s recorded singing. She had to feel comfortable with the recording: “It was Deborah’s job during the filming to mouth the approved playback properly. She would actually sing along, utilizing everything she had learned in our rehearsals. When you are lip-synching, there is no time to wait. The timing and impetus must be exact, not a second before or after.” During the actual filming later on, a technician was on hand to stop the filming if  Kerr’s lips were not in synchronization with Marni’s recording. But he rarely had to do this as most of the time Kerr was “flawless.” However, on a few occasions when things were changed from the original script during filming, Marni had to re-record a section. After this, her job was done.


West Side Story

Deborah Kerr had cooperated extremely well with Marni, but there was not such a good working relationship five years later in Marni’s next major project, West Side Story. The problem was that Natalie Wood had aspirations as a singer. She had a clause in her contract that she would record all her own songs, although she gave the producers the right to choose dubbing later. Wood was therefore uncomfortable having Marni, her potential dubber, present at her practice, but Marni had to be there to learn about Wood and the Puerto Rican accent she was using. The situation was very different from The King and I: “No working with the star and walking the set behind her this time.”



Wood’s situation was made more difficult in that her songs had complicated rhythms and “a high angular range.” And when she finally recorded, the orchestra showed its feelings by playing badly behind her and then playing well behind Marni and even applauding her. However, at this stage the producers still insisted that Wood’s voice would be used in the filming, mainly to keep her happy.


When the filming was completed and Wood had left, Marni was asked to dub all of Wood’s singing, something that the devious producers had planned all along. This meant using a different process: she would have to make a completely new recording “to fit my voice as Maria to her [Wood’s] lips on the screen.” To do this, the studio used a “loop” technique that allowed Marni to make multiple attempts to lip-synch the film, especially when the camera was in close-up on Wood’s lips. Marni recalled this work as “grueling,” but she managed to complete the work to her satisfaction.


She did have difficulties with the song “One Hand, One Heart.” She said later, “To sing it like a sixteen-year-old Puerto Rican girl, I had to keep it very, very light and un-operatic. The only way I could really manage it was to cry as I sang.”


My Fair Lady

When Marni was recruited to work on Audrey Hepburn’s songs in My Fair Lady, she had to cope again with an actress who had ambitions to be a singer. Fortunately, Hepburn was much more approachable than Wood, and though sensitive about her singing abilities, she made an effort to work with Marni. The two quickly became friends as Hepburn picked Marni up in her limo every morning. They had plenty of time to chat intimately on their way to work.


Hepburn allowed Marni to attend all her rehearsals and voice lessons: “Like a sponge, I was soaking up her unusual speech patterns, the way she breathed, the actual sound of her voice, which had a sexy semi-hoarse quality, while still being soft and feminine.”



Marni quickly felt “fused” with Hepburn. Together they wrestled with the different accents, especially Cockney. Hepburn’s insecurity really emerged once they both started to record the songs; she got increasingly frustrated with her efforts and often asked for an extra take. At this stage Marni didn’t know how much dubbing she would have to do. And Hepburn still believed that her singing voice would be used most of the time. As had happened with Natalie Wood, the producers maintained the pretense that this would happen, but as Marni found out later, “they never had any serious intention of using her voice.” Thus Marni had to record nearly all Hepburn’s songs, synchronizing with Hepburn’s lips from the film: “On several songs I recorded complete tracks, and on others I supplemented Audrey’s tracks with my higher notes.”


Dubbing Hepburn’s voice required all of Marni’s experience and skills: “Matching Audrey’s voice was a great challenge, much more difficult than matching that of either Deborah or Natalie. I had to be extra careful because her essential voice was not a soprano sound. Audrey seemed to have a wider, lower-shaped palate than mine, which was narrower and higher in shape. The anatomic differences can affect the high partials in the resonance and overtones in any voice. Thus, the very fabric of our tones was naturally different.”


Marni Nixon received little recognition for her work. In fact she was sworn to secrecy in the case of The King and I. She was poorly paid and was never granted royalties, except that for her work on West Side Story Leonard Bernstein gave her one quarter of one percent of his royalties—“not a negligible amount,” according to Marni. A total of 23 Oscars were awarded to the three major musical films that she worked on.


Note: The source for this article was Marni Nixon’s book, I Could Have Sung All Night (2006), which I highly recommend.


Sheila Dryden Tuesday 28th June 2016

Wow! fascinating. I loved all of these musicals and had no idea what a complicated process a stand in singer would necessitate.

Alan Buchanan Sunday 23rd April 2017

Fascinating story . Reminded me of the Student Prince where the intended star Mario Lanza was in no fit condition to play the lead as intended and little known actor Edmond Purdon was brought in using Lanza's dubbed vocals. The modern film Mamma Mia featured Hollywood A listers using their own voices to the detriment of Abba's songs. Marnie Nixon's voice will always exist on the sound track recordings of those musicals regardless of who appeared to be singing on screen.

John Parry Tuesday 22nd May 2018

I wish they had dubbed the voices in the film version of Les Miserables. What a wonderful play on the stage only to be ruined by the less than enjoyable movie voices.

Leave a Comment

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.