| |
Traditional |
Internet-based |
| Send us: |
DAT tape or computer audio files, plus the script. |
Computer audio files (.aiff, or .wav), plus the script. |
| Get back: |
Laser-printed paper output OR computer formats |
Various computer formats including postscript files
to print on your laser printer or other formats for use in software control of animation. |
| Shipped via: |
FedEx or UPS or email or internet |
email or internet |
| Processing time: |
One to three days, plus shipping time |
Seconds or minutes for each audio file |
| Availability: |
By arrangement. We try to be responsive. |
24 hours, 7 days. |
| Compared with other manual services: |
More flexible, more accurate, a little cheaper, not any faster (yet!) |
More flexible, more accurate in certain conditions, cheaper, faster. |
| Frame accurate information includes: |
Words, phonemes, line numbers, character names |
Words and phonemes only (until the special format for the script file is implemented in the automatic system). |
| X sheet page header/footer shows: |
Current footage (top and bottom of page) and sheet number;
you can also specify a title, production number, act and scene numbers, etc. |
Footage and sheet numbers numbers are automatically counted. The other items are not
yet implemented in the automatic system. |
| Video frame rates (30 fps or 29.997 fps) available? |
Yes. |
Yes. |
| Film frame rates (24 fps) available? |
Yes. |
Yes. |
| Can insert slugs into the X sheets? |
Yes. |
Not yet. |
| Mouth charting available? |
Experimentally, yes. |
Experimentally, yes. |
| Fits into your current production processes: |
At first this method will probably be easier to work with for traditional animation production. |
Either approach is easy to integrate into fully-computerized animation production processes. |
| Can it do overlapping voices? |
Yes; you simply specify a negative slug in the script, to specify the amount of overlap. |
Not yet |
| How is it done? |
An automatic first pass, followed by careful manual checking and correction. |
Proprietary, state of the art statistical signal processing
technology is used to automatically identify and locate speech sounds in your
audio tracks.
|
| Audio recording quality: |
Cleanly recorded mono audio track containing one voice only |
same |
| Performs well on: |
Speech or singing by adults or children in any English dialect or speaking style |
Standard American English adult speech or singing |
| Performs badly on: |
Overlapping voices, music, or background noise, which makes it
harder. That's why you should only submit clean, mono, voice-only recordings. |
Children's voices, distorted speech, some kinds of foreign accents. |
| Given "easy" recordings: |
Results are excellent for any kind of animation production |
Results are accurate enough for TV, but still needs checking for feature animation. |
| Given "hard" recordings: |
Results are excellent for any kind of animation production |
Results are not acceptable. |
| Will the system get better? |
Yes, both the automatic methods and the manual methods are
constantly being streamlined and improved. |
If you use either system, then the automatic system will learn
about your characters' voices, so eventually you will be able to use
the automatic system even for "hard" recordings. |
| Audio requirements: |
The DAT tapes should have the final takes of the voice recordings
for the entire show, with or without normal pauses or slugs inserted. |
The computer audio files should each be a recording of a single
script line per file; with or without reference tones -- assuming the
slugging is not done yet. If the audio is already slugged, then the
files should have a maximum length of up to 2 or 3 minutes.
|
| "As spoken" script: |
Presently use a clean ("plain ASCII") text file
Soon a special computer-processable format will
be available for you to specify slugging, line numbers, etc. |
We will convert your script to the special format for you |
| Free demo? |
By arrangement. |
Yes, any time. |
