1. TECHNICAL DESIGN
2. BASIC OPERATION
3. SEQUENCER MODE
4. TALK / SOUND
5. NOTE / SCALE
6. TREMOLO
7. RANDOM
8. MIDI SEQUENCER
9. EXPANDER MODE
10. MIDI CONTROLLER
11. HOW TO PRODUCE WORDS?


 

The internal sound is generated by the two independently controlled Speakjet ™ chips. Each of them provides a mono audio output (32 kHz PWM output) through a 6.3 mm stereo socket. The right audio channel is assigned to chip A, the left channel to chip B. An insert Y cable splits both channels into two mono audio outputs, allowing you to position the channels individually within the stereo sound field with any external mixing device.

The sound level ratio of both chips can be adjusted with the MIX control. An equal sound level is obtained by turning the MIX control to the centre position. All functions are clearly visible on the front panel with the start/ stop /MIDI-sync/ tempo functions on the upper right-hand corner, sound/ tremolo/ scale/ note/ pitch and random functions in the middle. The HOLD-switches are used to freeze the position of the joysticks.

The REC-switches are used to play back the recorded movements of the joysticks. You can switch between the modes SEQUENCER (SEQ) and EXPANDER (EXP) with the switch in the middle of the panel.

 

In SEQ mode sounds can be played manually. While the SEQUENCER is stopped (upper right switch is set to its lower position ‘STOP’) the TREMOLOS must be switched off (switches A and B in the TREMOLO section are set to their lower positions) in order not to interrupt the sound. The HOLD and REC switches should be in their upper positions.

The FLAME MIDI TALKING SYNTH is played by using the joystick TALK. Move the joystick out of its centre position to trigger any sounds (when the joystick is in the exact centre position, no sounds are triggered). As the HOLD function is inactive (i.e.: HOLD switch is in upper position), the joysticks movements are automatically recorded over a period of two measures. When you activate (i.e.: toggle down) the HOLD switch, the current position of the joystick will freeze and any further movements will have no effect.

Toggle down the REC switch to play back the recorded movements (any manual agitation of the joystick has no effect now). The tempo of the play-back depends on the internal respectively external MIDI tempo. The MIDI tempo is signalled by the SCALE LED, which flashes at the beginning of each measure serving as a visual control mechanism for the recording period.

The sounds you play depend on their note values (position of the joystick, scale, SCALE-switch). Moving the joystick vertically alters the pitch within a range of 3 octaves. Moving the joystick horizontally changes the formants of the sound (male-female) giving it the typical robot-like sound effect. Use the HOLD and REC switches to freeze and play-back the sounds you play.

Once the sequencer has been started (internally or externally), you can interrupt the audio signal with the TREMOLO by toggling up one or both TREMOLO switches. The LEDs indicate the rhythm (flashing = audio on). This can be set individually for each Speakjet ™ chips. The 8/16 switch enables you to switch the quantization between quavers and semi quavers. The next switch to the right sets up the phase. Both settings apply for A and B.

It is not possible to trigger sounds in the SEQ mode via external MIDI note.

In the EXP mode the synthesizer works just like a drum expander module. Words and sounds are triggered only by received MIDI notes. No sequencer/ tremolo/ random or scale functions are available in this mode.

 

As outlined above in this mode sounds are played manually and are interrupted by the MIDI clock-controlled TREMOLO.
Sounds can also be played without sequencer running. The sequencer is controlled by the internal or external MIDI clock and sends MIDI TIMING data (start/ stop/ clock) when running.

 

joystick TALK
This is the control that triggers the sound. No sound is triggered in the centre position unless the corresponding switch HOLD or REC are in the lower position and generate frozen or recorded joystick data.
64 sounds can be played. They are arranged in 8 lists each containing 16 sound elements. Each sound bank contains different lists. Above the centre position the sounds of the lists are played individually. Below the centre position the sounds of the lists are played in sequence depending on the joystick’s position.
This way short sequences can be played in loops, which consist of up to 16 elements. Some of the 12 sound banks contain fewer elements so that it is easier to find the same sounds.

HOLD switch (Talk)
This switch freezes the current position of the joystick. Only those sounds are played back.
Please note: No sounds are played while the joystick is in centre position.

REC switch (Talk)
With this switch the joystick movement is continually recorded over the period of two measures. When the switch is toggled down, the recorded movements are played back. Manual playing is not possible then.
Please note that the tempo depends on the MIDI tempo. The LED flashes for every bar signalling the recording window (when sequencer is stopped). Recording as well as playing back lasts two bars and is then continually repeated. When the sequencer is started the MIDI clock LEDs give a visual control.

LOOP switch
In the lower position (LOOP on) any chosen sound is repeated continually. In the upper position (LOOP off) the sound elements are only played once. This applies for Speakjet A and B.

SOUND turn-switch
12 patches are available. You can switch between them at runtime. There are two sound banks for each patch for each Speakjet which are triggered with the SOUND TALK/ SYNTH switch.
SOUNDS TALK: lists with allophones that can be played tonally
SOUNDS SYNTH: lists with effect sounds that cannot be played tonally
See appendix for an overview of all patches.

SPEED control
Playback velocity of the allophones: left = slow, right = fast
Please note:SPEED and TEMPO should not be mixed up. The TEMPO control regulates the playback tempo of the internal sequencer with effect on the tremolo while the SPEED control changes the duration of the generated talk/ synth sounds (not MIDI synchronized).

MIX control
MIX control controls the volume for Speakjet A and B. In the centre position they have the same volume
Please note: this control is digital. Delays are possible as the data has to be processed in the internal Speakjet buffer (FIFO). This can affect different SPEED settings.

 

joystick NOTE
Horizontal movements affect the sound by changing the formants (male/ female), vertical movements alter the pitch (depending on the setting of the SCALE filter). When SCALE is switched off, all semitones can be played within a range of 3 octaves. When SCALE is switched on, the notes of a set scale are played.

SCALE switch
ON/OFF - activates/ deactivates the key filter
DUR/MOLL - toggles between major key (DUR) and harmonic minor key (MOLL)
DREHSCHALTER - chooses the key


 

TREMOLO only works when the sequencer is started with internal/ external MIDI clock/ tempo. It can be switched on for each Speakjet individually.
When both are switched on the phase can be shifted with the PHASE switch.
The 8/16 switch determines the beat. Visual control is provided by two LEDs.

 

A number of random algorithms are implemented in the device that can be triggered by different positions of the RANDOM switch for each Speakjet (A and B). There is no random effect in the centre position as different algorithms work in the right and in the left position. Starting from the centre position various effects can be triggered.

Moving to the left affects first pitch, then speed, then volume. It is most intense in the very left position. Any random variations start from the positions of the NOTE/ SPEED/ MIX controls.

The changes are not always linear and do not cover the whole range of all values, so experimenting can have interesting results.

 

To use the internal sequencer the left switch above the TEMPO control must be set to INT (internal MIDI clock). Now the sequencer can be started and stopped with the START/ STOP switch. In this modus the device works as a MIDI master and externally connected devices can run synchronously via the MIDI clock. The tempo is controlled by the TEMPO control. The control range extends from tempo 20 to 220. In the centre position the approximate tempo is 115. Please note that not all tempi can be set by the control due to the limited resolution.

To use the sequencer as a slave the SYNC switch must be set to EXT (external MIDI clock). Now the sequencer can be controlled by a external sequencer via the MIDI clock (START/ STOP/ CLOCK). Please note: The device does not understand CONTINUE. The TEMPO control has no effect on the tempo played. It continues to send as a MIDI controller and can be used for other tasks as well.

Please note: When the sequencer is running with the external clock you can switch to INT during normal operation and leave it running with the current tempo setting “out of time” (right switch must be on START). If you want to switch back to EXT it runs in sync with the external MIDI clock as it continues to register the external clock. Rhythmic variations can be tried this way during normal operation.

The red LED signals the first beat of a bar, the yellow one the quarter note.

The MIDI sequencer (INT and EXT) always sends the MIDI clock data via MIDI OUT. It does not send NOTE ON/ OFF data.

Attention: Should problems with the timing arise with the external synchronisation, it should be checked if the device receives more than just MIDI clock data. Additional data such as ProgramChange, controlChange, SysEX- or note data should be avoided or filtered out!

 

In this mode the device works like a sound expander, like a drum module in fact. This means that certain sounds and whole words are assigned to different keys and only those can be played on the keyboard. The assigned words and sounds are listed in the attachment. The device is set on MIDI channel 10 and the setting cannot be changed. The sounds also can be changed with the SPEED control and with the joystick NOTE. Joystick HOLD can be used as well. Turn the SOUND switch to select patches. Any other functions are inactive. Only the sending of MIDI control data is the same as in the SEQ mode.

Please note: Speakjets A and B generate the same sounds in this mode.
Attention: The MIX control should be in the centre position before switching from SEQ to EXP mode to hear both Speakjets.

The position of the TREMOLO switch has no effect on the audio output in this modus. It is always switched on. The red scale LED signals received MIDI notes.

 

All controls send MIDI control-change data during runtime. Each control sends on both MIDI channels 1 and 2 using the same control change number. See the appendix for further information on which numbers are assigned to which control.

Please note: In SEQ mode data-sending is linked to the tick of the internal or external MIDI clock and therefore depends on the given tempo.

In the EXP mode data is sent faster as there is no MIDI clock. Also two identical data values are never sent in succession in order to reduce the data flow — only when the data value of a controller changes new data is sent.

 

At first:
The possibility to create words with the Talking-Synth is only limited. Use the phonemes/allophones from patch 6 with an external sequencer.
Please note: it's very difficult to build words manually only with a keyboard.

Send the sounds, which build a word, one after another at intervals of 1 MIDI Tick (duration/gatetime = 1 Tick). Set the song tempo on 40..50 bpm. Set the speed controller from Talking-Synth between middle and max position. If the speed onto min position, then set the song tempo on 20 bpm. By variations of speed and tone pitch its possibble to improve the understandability. Its helpful to experience with different adjustments and sounds.

For example: Load the Standard-MIDI-File „Flame_Talking_Synth.mid“ in your sequencer and send the track data (note numbers) on Midi channel 10 to the Talking-Synth (settings: Expander mode, Speed on max position). Now listen: "Flame MIDI-Talking-Synth Speech-Synthesizer"
Please note: Send from sequencer to the Talking-Synth exclusively the sound note numbers ( not any datas).

Please note: The sounds in patch 6 are very different loud. Furthermore some phonemes are different in sound and are not identical with the sound of letter. A multiplicity of combinations generate the effect of letters in words.

Five basic linguistic concepts will help you to create highly intelligible sounding speech:

1. there is no one-to-one correspondence between written letters and speech sounds. More than one letter may represent each sound in a language and, conversely each letter may represent more than one sound. Because of these spelling irregularities, it is necessary to think in terms of sounds, not letters, when creating phrases.

2. speech sounds are acoustically different depending upon their position within a word. For example the initial B sound in "Beep" is acoustically different from the B sound in "Box" The B sounds differ due to the influence of the vowels, which follow them.

3. the human ear may perceive the same acoustic signal differently in the context of different sounds. Therefore, an allophone may sound slightly different when used in various phrases.

4. some sounds in words are not actually pronounced and/or others are added when followed or preceded by certain other words. For example the two words “Night” and “Time”. When played separately, they both will articulate the “T” sound. However, played together they need to be articulated as “Ni-Time”.

5. The Vocalization Pitch, Play Rate and Frequency Bend parameters are just as important as the selection of the phonemes used. For instance, playing a phase that does not change the pitch at any point in time sounds very monotonic and robotic, while increasing the Pitch at the end of a sentence produces a questioning tone.