SYSTEM AND METHOD FOR SIMULTANEOUS MULTILINGUAL DUBBING

OF VIDEO-AUDIO PROGRAMS

CROSS REFERENCE TO RELATED APPLICATIONS

This Present Application is the non-provisional counterpart of U.S. Provisional Patent Application Serial No. 62/814,419 filed on March 6, 2019. This Present Application claims the benefit and priority of said Provisional Patent Application, which is incorporated by reference in its entirety herein. BACKGROUND OF THE INVENTION

Dubbing of videos with dialog in a language different from that of the active speakers is difficult and cumbersome. It is often done by human voice over in order to synchronize the translated speech patterns to the speaker’s lip movements and to closely reproduce the speaker’s original intonation and emphasis. Good voice actors are expensive, and Oscar-winning voice actors simply do not exist.

Each dubbing exercise for a particular language is a separate activity. Dubbing of a video in Spanish is different and separate from dubbing of the same video in French. Moreover, close captioned subtitles are required in addition to the voice-over. For example, an English video for the Chinese market requires dubbing into Mandarin. However, Mandarin is only one of several Chinese languages. Nevertheless, although the Chinese spoken languages are different, they are written identically. Thus, a Cantonese and Mandarin speaker can read and understand the same written text. To that end, almost all Chinese videos have Chinese subtitles.

Thus, there is a need for a system and method for automatically performing multilingual dubbing of videos. This would make such dubbing far less expensive. It would replace humans for all but the highest budget dubbing projects.

SUMMARY OF THE INVENTION

The Present Invention is for a system and method to perform dubbing automatically for multiple languages at the same time using speech-to-text transcriptions, language translation, and artificial intelligence engines to perform the actual dubbing in the voice likeness of the original speaker. While speech-to-text, machine language translation, and text-to-speech conversion hardware and software are state-of-art, their combination to produce the Present Invention is novel and unique.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic representation of the process steps for producing a dubbing project.

FIG. 2A illustrates how the speech-to-text component of the Present Invention distinguishes between words, phrases, and sentences based upon pauses between words.

FIB. 2B describes parameters associated with each word, phrase, and sentence that is collected and stored when automatically analyzing a source audio stream.

DETAILED DESCRIPTION OF THE INVENTION

There are two primary embodiments of the Present Invention. One embodiment produces the dubbed video as a real time audio/video stream, and the other embodiment creates the finished product offline. Referring to FIG. 1 , in an exemplary embodiment, a video program containing both audio and video is transmitted from a video source 1 via element 2 to a device 3 that processes the program and transmits it via element 4 to a cloud network 5. As referred to herein, the video program is a series of video frames containing an audio track. Device 3 is a cloud interface capable of receiving the unprocessed video from source 1 or transmitting the finished video back to source 1. For example without limitation, the video source 1 may be a broadcast station or a web streaming service. In another embodiment, cloud processing could be eliminated. A local area network, wide area network, or the Internet would be a reasonable substitution. Further, there could be a direct connection via element 6 from element 2 to transcription service 7, which is on a separate computer.

In all embodiments, the system of the Present Invention transmits the video program via element 6 to transcription service 7, which produces a text script of the audio program in the originally recorded language using a speech-to-text engine 8. A computerized or human transcription may be used. Text-to-speech software recognizes phonemes, and it uses a dictionary to form words. The computerized engine 8 uses artificial intelligence to distinguish between various speakers and to assign the text strings to those speakers. Further, the system also transcribes and synchronizes inflection, emphasis, and volume variations to the text. The system is capable of distinguishing between male and female speakers (including children), and it assigns these identification parameters to the text. The identification parameters could include a "raspness" index to add character to the voice. A synchronizer 9 automatically attaches timing parameters to each word in the text string. These timing parameters measure the temporal length of each word and synchronize the inflection, emphasis, and volume indicators with various temporal points within each string.

The timing parameters establish the start time and the end time for each word. In this way, the transcription algorithm can measure the temporal length of pauses in speech. FIG. 2A illustrates how timing of pauses is used to analyze the text. The shortest pauses are between words continuously strung into a phrase. There are longer pauses between phrases, and even longer pauses between sentences. Very long pauses indicate that there is no speech to transcribe. Thus, an audio stream may be transcribed into words that are grouped in phrases and sentences. Sentences consist of one or more phrases. The parmeters to be collected are shown in FIG. 2B. In any phrase, emphasis will invariably be on its last word. That word will have greater intensity of volume. This is important when it comes to language translation. Within a given phrase, there will be some words having greater relative volume than the other words, and the last word having greater relative volume than the others. Emphasis is established using relative volume.

Any given phrase will be spoken by the same person. Thus the parameters of gender and age will be constant within the phrase. With rare exceptions, this will also apply to sentences.

An artificial intelligence component of the software determines the emotional aspect of each phrase or sentence. This is determined by the way words are uttered in sequence. People often sing when they speak. Software can detect when a person is whining by the tonality of words, their location in a phrase or sentence, and how fast the words are uttered relative to each other. The software is able to detect when speakers are happy, sad, frightened, etc. FIG. 2B shows many of the parameters associated with words and phrases. The text strings are simultaneously translated phrase by phrase into multiple languages by translation engine 10. The system then produces multiple scripts each comprising a series of concatenated text strings representing phrases along with associated inflection, emphasis, volume, and emotional indicators as well as timing and speaker identifiers that are derived from the original audio signal. Each text string in both the untranslated and translated versions has a series of timing points. The system synchronizes these timing points of the words and phrases of the translated strings to those of the untranslated strings. It is important that the translated string retains the emotional character of the original source. Thus, intonations of certain words and phrases in both the translated and source text strings is retained along with volume, emphasis, and relative pause lengths within the strings.

Within a phrase, the number and order of words might be different for different languages. This is based on grammar discrepancies in different languages. For example, in German, verbs normally appear at the end of a phrase, as opposed to English where subjects and verbs maintain close proximity. Single words could translate to multiple words and vice versa. For example, in many languages, a potato is an earth apple. In French, this translation has the same number of syllables, but in other languages, there could be more or less syllables. That is why it is difficult to translate songs from language to another while keeping the same melody. In any event, the beginning and end temporal points for each phrase must be the same in the original source text and the translated target text. Thus, when translated voice dubbing occurs, speech cadence in the dubbed translation may be sped up or slowed down so that temporal beginning and end points of any phrase would be the same in any language.

Voice dubbings are created from the text strings using a text-to-speech module. All of the parameters contained in the text strings associated with each word, phrase, and sentence are used to create the audio stream. Thus, speech made by a person in the target language will sound exactly like the speech made by the same person in the source language. All of the voice and emotional characteristics will be retained for each person in each phrase. It will appear as if the same speaker is talking in a different language. Multiple language dubbings are simultaneously produced for all translated scripts using dubbing engine 11. Here, text-to-speech synthesizers are used to create audio strings in various languages, corresponding to phrases, that are time synchronized to their original language audio strings. Corresponding translated words are given the same relative volume and emphasis indicators as their source counterparts. Each audio string has multiple temporal points that correspond to those in their respective text strings. In this way, the translated language strings fully correspond in time to the original language strings. Various speakers are assigned individual voiceprints based on sex, age and other factors. The intonation, emphasis, and volume indicators ensure that the voice dubbings sound realistic and as close to the original speaker's voice as possible.

Close captioning (CC) is another factor to consider. Where this is desired, the translated text is either flashed or scrolled onto the screen as subtitles. The system has the ability to determine the placement of the subtitles on the screen so as not to interfere with the focus of the video program content.

An Analysis Module 12 analyzes the placement and superposition of the subtitles onto the original video program. Once this has been done (using artificial intelligence), the dubbed video is sent back to the cloud via element 14, and then back to video source 1 via element 15.

The real time embodiment requires an extra step, i.e., Step 13, where transmission of the video program back to video source 1 is delayed to allow synchronization of the dubbed audio to the video. The delay is very short, being a fraction of a minute.

The offline or non-real time embodiment functions similarly to the real time embodiment except that more humans may be added into the loop to effect cleanup and quality control. The primary difference is that the offline embodiment provides more accuracy due to human intervention. The following represents some of the workflow differences that may occur with the offline embodiment.

1. Humans may transcribe the audio rather than relying on a machine transcription.

2. The transcription may better be synchronized with the speech. 3. There is more opportunity for quality control.

4. Human language translation is often more accurate and localized than machine language translation.

5. A graphical user interface (GUI) interface may be used to edit the synthetic dubbed audio for the following features:

a. audio volume - loudness or softness;

b. compression of the words to comply with the rate of speech; and c. intonation - emphasis of the words and voice can be adjusted to be the same as in the originally recorded speech.

Other cleanup tools would allow:

• editing speech-to-text;

• editing timing;

• editing diarization; and

• editing the prosody/intonation, voice, and other aspects of generated speech.