echogarden 1.3.2 → 1.4.0

package/dist/utilities/LEB128.js

@@ -0,0 +1,168 @@
+import { logToStderr } from "./Utilities.js";
+////////////////////////////////////////////////////////////////////////////////////
+// Encode
+////////////////////////////////////////////////////////////////////////////////////
+export function encodeSignedInt32(value, outEncodedData) {
+    if (value < -2147483648 || value > 2147483647) {
+        throw new Error('Value must be between -2147483648 and 2147483647');
+    }
+    while (true) {
+        const lowest7Bits = value & 127;
+        value >>= 7;
+        if ((value === 0 && (lowest7Bits & 64) === 0) ||
+            (value === -1 && (lowest7Bits & 64) !== 0)) {
+            outEncodedData.push(lowest7Bits);
+            return outEncodedData;
+        }
+        else {
+            outEncodedData.push(lowest7Bits | 128);
+        }
+    }
+}
+export function encodeSignedInt32sFast(value, outEncodedData) {
+    const absValue = Math.abs(value);
+    //const absMask = value >> 31
+    //const absValue = (value ^ absMask) - absMask
+    if (absValue < (2 ** 6)) {
+        outEncodedData.push((value & 127));
+    }
+    else if (absValue < (2 ** 13)) {
+        outEncodedData.push((value & 127) | 128, (value >> 7) & 127);
+    }
+    else if (absValue < (2 ** 20)) {
+        outEncodedData.push((value & 127) | 128, ((value >> 7) & 127) | 128, (value >> 14) & 127);
+    }
+    else if (absValue < (2 ** 27)) {
+        outEncodedData.push((value & 127) | 128, ((value >> 7) & 127) | 128, ((value >> 14) & 127) | 128, (value >> 21) & 127);
+    }
+    else if (value < (2 ** 31) && value >= -(2 ** 31)) {
+        outEncodedData.push((value & 127) | 128, ((value >> 7) & 127) | 128, ((value >> 14) & 127) | 128, ((value >> 21) & 127) | 128, (value >> 28) & 127);
+    }
+    else {
+        throw new Error(`Value must be between -2147483648 and 2147483647`);
+    }
+}
+////////////////////////////////////////////////////////////////////////////////////
+// Decode
+////////////////////////////////////////////////////////////////////////////////////
+export function decodeSignedInt32s(encodedData, outDecodedValues) {
+    for (let readIndex = 0; readIndex < encodedData.length;) {
+        let currentDecodedValue = 0;
+        let shiftAmount = 0;
+        while (true) {
+            const encodedByte = encodedData[readIndex++];
+            const lowest7Bits = encodedByte & 127;
+            currentDecodedValue |= lowest7Bits << shiftAmount;
+            // If 8th bit is 0, then this is the last byte in the sequence
+            if ((encodedByte & 128) === 0) {
+                // If 7th bit is 1, then the value is negative
+                if ((encodedByte & 64) !== 0) {
+                    // If the value should be negative
+                    // Ensure that the value is encoded as a negative number by
+                    // setting all higher bits to 1
+                    currentDecodedValue |= -1 << Math.min(shiftAmount + 7, 31);
+                }
+                break;
+            }
+            if (readIndex === encodedData.length) {
+                throw new Error(`Invalid LEB128 data. Last encoded byte sequence is truncated.`);
+            }
+            shiftAmount += 7;
+            if (shiftAmount > 31) {
+                throw new Error(`LEB128 sequence can't be decoded. Byte sequence extends beyond the range of a signed 32 bit integer.`);
+            }
+        }
+        outDecodedValues.push(currentDecodedValue);
+    }
+    return outDecodedValues;
+}
+export function decodeSignedInt32sFast(encodedData, outDecodedValues) {
+    for (let readIndex = 0; readIndex < encodedData.length;) {
+        const byte0 = encodedData[readIndex++];
+        if ((byte0 & 128) === 0) {
+            let decodedValue = (byte0 & 127);
+            if ((byte0 & 64) !== 0) {
+                decodedValue |= -1 << 7;
+            }
+            outDecodedValues.push(decodedValue);
+            continue;
+        }
+        const byte1 = encodedData[readIndex++];
+        if ((byte1 & 128) === 0) {
+            let decodedValue = (byte0 & 127) |
+                (byte1 & 127) << 7;
+            if ((byte1 & 64) !== 0) {
+                decodedValue |= -1 << 14;
+            }
+            outDecodedValues.push(decodedValue);
+            continue;
+        }
+        const byte2 = encodedData[readIndex++];
+        if ((byte2 & 128) === 0) {
+            let decodedValue = (byte0 & 127) |
+                (byte1 & 127) << 7 |
+                (byte2 & 127) << 14;
+            if ((byte2 & 64) !== 0) {
+                decodedValue |= -1 << 21;
+            }
+            outDecodedValues.push(decodedValue);
+            continue;
+        }
+        const byte3 = encodedData[readIndex++];
+        if ((byte3 & 128) === 0) {
+            let decodedValue = (byte0 & 127) |
+                (byte1 & 127) << 7 |
+                (byte2 & 127) << 14 |
+                (byte3 & 127) << 21;
+            if ((byte3 & 64) !== 0) {
+                decodedValue |= -1 << 28;
+            }
+            outDecodedValues.push(decodedValue);
+            continue;
+        }
+        const byte4 = encodedData[readIndex++];
+        if ((byte4 & 128) === 0) {
+            let decodedValue = (byte0 & 127) |
+                (byte1 & 127) << 7 |
+                (byte2 & 127) << 14 |
+                (byte3 & 127) << 21 |
+                (byte4 & 127) << 28;
+            if ((byte4 & 64) !== 0) {
+                decodedValue |= -1 << 31;
+            }
+            outDecodedValues.push(decodedValue);
+            continue;
+        }
+        if (readIndex >= encodedData.length) {
+            throw new Error(`Invalid LEB128 data. Last encoded byte sequence is truncated.`);
+        }
+        else {
+            throw new Error(`LEB128 sequence can't be decoded. Encoded byte sequence represents a value that extends beyond the range of a signed 32 bit integer.`);
+        }
+    }
+    return outDecodedValues;
+}
+////////////////////////////////////////////////////////////////////////////////////
+// Tests
+////////////////////////////////////////////////////////////////////////////////////
+export function testLeb128() {
+    const encodedBytes = [];
+    const decodedValues = [];
+    function runTest(testValue) {
+        encodedBytes.length = 0;
+        decodedValues.length = 0;
+        encodeSignedInt32sFast(testValue, encodedBytes);
+        decodeSignedInt32sFast(encodedBytes, decodedValues);
+        if (decodedValues[0] !== testValue) {
+            throw new Error(`Expected ${testValue} but got ${decodedValues[0]}`);
+        }
+    }
+    for (let i = -(2 ** 22); i < 2 ** 22; i++) {
+        if (i % 1000000 === 0) {
+            logToStderr(i);
+        }
+        runTest(i);
+    }
+    const x = 1;
+}
+//# sourceMappingURL=LEB128.js.map

package/dist/utilities/LEB128.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"LEB128.js","sourceRoot":"","sources":["../../src/utilities/LEB128.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,gBAAgB,CAAA;AAE5C,oFAAoF;AACpF,SAAS;AACT,oFAAoF;AACpF,MAAM,UAAU,iBAAiB,CAAC,KAAa,EAAE,cAAwB;IACxE,IAAI,KAAK,GAAG,CAAC,UAAU,IAAI,KAAK,GAAG,UAAU,EAAE,CAAC;QAC/C,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC,CAAA;IACpE,CAAC;IAED,OAAO,IAAI,EAAE,CAAC;QACb,MAAM,WAAW,GAAG,KAAK,GAAG,GAAG,CAAA;QAE/B,KAAK,KAAK,CAAC,CAAA;QAEX,IACC,CAAC,KAAK,KAAK,CAAC,IAAI,CAAC,WAAW,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;YACzC,CAAC,KAAK,KAAK,CAAC,CAAC,IAAI,CAAC,WAAW,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;YAC7C,cAAc,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;YAEhC,OAAO,cAAc,CAAA;QACtB,CAAC;aAAM,CAAC;YACP,cAAc,CAAC,IAAI,CAAC,WAAW,GAAG,GAAG,CAAC,CAAA;QACvC,CAAC;IACF,CAAC;AACF,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,KAAa,EAAE,cAAwB;IAC7E,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;IAChC,6BAA6B;IAC7B,8CAA8C;IAE9C,IAAI,QAAQ,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACzB,cAAc,CAAC,IAAI,CAClB,CAAC,KAAK,GAAG,GAAG,CAAC,CACb,CAAA;IACF,CAAC;SAAM,IAAI,QAAQ,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QACjC,cAAc,CAAC,IAAI,CAClB,CAAC,KAAK,GAAG,GAAG,CAAC,GAAG,GAAG,EACnB,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,GAAG,CAClB,CAAA;IACF,CAAC;SAAM,IAAI,QAAQ,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QACjC,cAAc,CAAC,IAAI,CAClB,CAAC,KAAK,GAAG,GAAG,CAAC,GAAG,GAAG,EACnB,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,GAAG,GAAG,EAC1B,CAAC,KAAK,IAAI,EAAE,CAAC,GAAG,GAAG,CACnB,CAAA;IACF,CAAC;SAAM,IAAI,QAAQ,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QACjC,cAAc,CAAC,IAAI,CAClB,CAAC,KAAK,GAAG,GAAG,CAAC,GAAG,GAAG,EACnB,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,GAAG,GAAG,EAC1B,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC,GAAG,GAAG,CAAC,GAAG,GAAG,EAC3B,CAAC,KAAK,IAAI,EAAE,CAAC,GAAG,GAAG,CACnB,CAAA;IACF,CAAC;SAAM,IAAI,KAAK,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;QACrD,cAAc,CAAC,IAAI,CAClB,CAAC,KAAK,GAAG,GAAG,CAAC,GAAG,GAAG,EACnB,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,GAAG,GAAG,EAC1B,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC,GAAG,GAAG,CAAC,GAAG,GAAG,EAC3B,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC,GAAG,GAAG,CAAC,GAAG,GAAG,EAC3B,CAAC,KAAK,IAAI,EAAE,CAAC,GAAG,GAAG,CACnB,CAAA;IACF,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC,CAAA;IACpE,CAAC;AACF,CAAC;AAED,oFAAoF;AACpF,SAAS;AACT,oFAAoF;AACpF,MAAM,UAAU,kBAAkB,CAAC,WAA8B,EAAE,gBAA0B;IAC5F,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,WAAW,CAAC,MAAM,GAAG,CAAC;QACzD,IAAI,mBAAmB,GAAG,CAAC,CAAA;QAC3B,IAAI,WAAW,GAAG,CAAC,CAAA;QAEnB,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,WAAW,GAAG,WAAW,CAAC,SAAS,EAAE,CAAC,CAAA;YAC5C,MAAM,WAAW,GAAG,WAAW,GAAG,GAAG,CAAA;YAErC,mBAAmB,IAAI,WAAW,IAAI,WAAW,CAAA;YAEjD,8DAA8D;YAC9D,IAAI,CAAC,WAAW,GAAG,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;gBAC/B,8CAA8C;gBAC9C,IAAI,CAAC,WAAW,GAAG,EAAE,CAAC,KAAK,CAAC,EAAE,CAAC;oBAC9B,kCAAkC;oBAClC,2DAA2D;oBAC3D,+BAA+B;oBAC/B,mBAAmB,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,GAAG,CAAC,WAAW,GAAG,CAAC,EAAE,EAAE,CAAC,CAAA;gBAC3D,CAAC;gBAED,MAAK;YACN,CAAC;YAED,IAAI,SAAS,KAAK,WAAW,CAAC,MAAM,EAAE,CAAC;gBACtC,MAAM,IAAI,KAAK,CAAC,+DAA+D,CAAC,CAAA;YACjF,CAAC;YAED,WAAW,IAAI,CAAC,CAAA;YAEhB,IAAI,WAAW,GAAG,EAAE,EAAE,CAAC;gBACtB,MAAM,IAAI,KAAK,CAAC,sGAAsG,CAAC,CAAA;YACxH,CAAC;QACF,CAAC;QAED,gBAAgB,CAAC,IAAI,CAAC,mBAAmB,CAAC,CAAA;IAC3C,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,WAA8B,EAAE,gBAA0B;IAChG,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,WAAW,CAAC,MAAM,GAAG,CAAC;QACzD,MAAM,KAAK,GAAG,WAAW,CAAC,SAAS,EAAE,CAAC,CAAA;QAEtC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;YACzB,IAAI,YAAY,GACf,CAAC,KAAK,GAAG,GAAG,CAAC,CAAA;YAEd,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC,EAAE,CAAC;gBACxB,YAAY,IAAI,CAAC,CAAC,IAAI,CAAC,CAAA;YACxB,CAAC;YAED,gBAAgB,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;YAEnC,SAAQ;QACT,CAAC;QAED,MAAM,KAAK,GAAG,WAAW,CAAC,SAAS,EAAE,CAAC,CAAA;QAEtC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;YACzB,IAAI,YAAY,GACf,CAAC,KAAK,GAAG,GAAG,CAAC;gBACb,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC,CAAA;YAEnB,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC,EAAE,CAAC;gBACxB,YAAY,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;YACzB,CAAC;YAED,gBAAgB,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;YAEnC,SAAQ;QACT,CAAC;QAED,MAAM,KAAK,GAAG,WAAW,CAAC,SAAS,EAAE,CAAC,CAAA;QAEtC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;YACzB,IAAI,YAAY,GACf,CAAC,KAAK,GAAG,GAAG,CAAC;gBACb,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC;gBAClB,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE,CAAA;YAEpB,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC,EAAE,CAAC;gBACxB,YAAY,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;YACzB,CAAC;YAED,gBAAgB,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;YAEnC,SAAQ;QACT,CAAC;QAED,MAAM,KAAK,GAAG,WAAW,CAAC,SAAS,EAAE,CAAC,CAAA;QAEtC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;YACzB,IAAI,YAAY,GACf,CAAC,KAAK,GAAG,GAAG,CAAC;gBACb,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC;gBAClB,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE;gBACnB,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE,CAAA;YAEpB,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC,EAAE,CAAC;gBACxB,YAAY,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;YACzB,CAAC;YAED,gBAAgB,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;YAEnC,SAAQ;QACT,CAAC;QAED,MAAM,KAAK,GAAG,WAAW,CAAC,SAAS,EAAE,CAAC,CAAA;QAEtC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;YACzB,IAAI,YAAY,GACf,CAAC,KAAK,GAAG,GAAG,CAAC;gBACb,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC;gBAClB,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE;gBACnB,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE;gBACnB,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE,CAAA;YAEpB,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC,EAAE,CAAC;gBACxB,YAAY,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;YACzB,CAAC;YAED,gBAAgB,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;YAEnC,SAAQ;QACT,CAAC;QAED,IAAI,SAAS,IAAI,WAAW,CAAC,MAAM,EAAE,CAAC;YACrC,MAAM,IAAI,KAAK,CAAC,+DAA+D,CAAC,CAAA;QACjF,CAAC;aAAM,CAAC;YACP,MAAM,IAAI,KAAK,CAAC,sIAAsI,CAAC,CAAA;QACxJ,CAAC;IACF,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,oFAAoF;AACpF,QAAQ;AACR,oFAAoF;AACpF,MAAM,UAAU,UAAU;IACzB,MAAM,YAAY,GAAa,EAAE,CAAA;IACjC,MAAM,aAAa,GAAa,EAAE,CAAA;IAElC,SAAS,OAAO,CAAC,SAAiB;QACjC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAA;QACvB,aAAa,CAAC,MAAM,GAAG,CAAC,CAAA;QAExB,sBAAsB,CAAC,SAAS,EAAE,YAAY,CAAC,CAAA;QAC/C,sBAAsB,CAAC,YAAY,EAAE,aAAa,CAAC,CAAA;QAEnD,IAAI,aAAa,CAAC,CAAC,CAAC,KAAK,SAAS,EAAE,CAAC;YACpC,MAAM,IAAI,KAAK,CAAC,YAAY,SAAS,YAAY,aAAa,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;QACrE,CAAC;IACF,CAAC;IAED,KAAK,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE,CAAC,EAAE,EAAE,CAAC;QAC3C,IAAI,CAAC,GAAG,OAAO,KAAK,CAAC,EAAE,CAAC;YACvB,WAAW,CAAC,CAAC,CAAC,CAAA;QACf,CAAC;QAED,OAAO,CAAC,CAAC,CAAC,CAAA;IACX,CAAC;IAED,MAAM,CAAC,GAAG,CAAC,CAAA;AACZ,CAAC"}

package/docs/Options.md

@@ -149,6 +149,7 @@ Applies to CLI operation: `transcribe`, API method: `recognize`
 * `whisper.autoPromptParts`: use previous part's recognized text as the prompt for the next part. Disabling this may help to prevent repetition carrying over between parts, in some cases. Defaults to `true`
 * `whisper.maxTokensPerPart`: maximum number of tokens to decode for each audio part. Defaults to `250`
 * `whisper.suppressRepetition`: attempt to suppress decoding of repeating token patterns. Defaults to `true`
+* `whisper.repetitionThreshold`: minimal repetition / compressibility score to cause a part not to be auto-prompted to the next part. Defaults to `2.4`
 * `whisper.decodeTimestampTokens`: enable/disable decoding of timestamp tokens. Setting to `false` can reduce the occurrence of hallucinations and token repetition loops, possibly due to the overall reduction in the number of tokens decoded. This has no impact on the accuracy of timestamps, since they are derived independently using cross-attention weights. However, there are cases where this can cause the model to end a part prematurely, especially in singing and less speech-like voice segments, or when there are multiple speakers. Defaults to `true`
 * `whisper.encoderProvider`: identifier for the ONNX execution provider to use with the encoder model. Can be `cpu` or `dml` ([DirectML](https://microsoft.github.io/DirectML/)-based GPU acceleration - Windows only). In general, GPU-based encoding should be significantly faster. Defaults to `cpu`, or `dml` if available
 * `whisper.decoderProvider`: identifier for the ONNX execution provider to use with the decoder model. Can be `cpu` or `dml` (Windows only). Using GPU acceleration for the decoder may be faster than CPU, especially for larger models, but that depends on your particular combination of CPU and GPU. Defaults to `cpu`
@@ -221,7 +222,7 @@ Applies to CLI operation: `align`, API method: `align`
 * `plainText.whitespace`: determines how to process whitespace within transcript paragraphs. Can be `preserve` (leave as is), `removeLineBreaks` (convert line breaks to spaces) or `collapse` (convert runs of whitespace characters, including line breaks, to a single space character). Defaults to `collapse`
 
 **DTW**:
-* `dtw.granularity`: adjusts the MFCC frame width and hop size based on the profile selected. Can be set to either `auto` (auto-selected based on audio duration and task), `xx-low` (400ms width, 160ms hop), `x-low` (200ms width, 80ms hop), `low` (100ms width, 40ms hop), `medium` (50ms width, 20ms hop), `high` (25ms width, 10ms hop), `x-high` (20ms width, 5ms hop). For multi-pass processing, multiple granularities can be provided, like `dtw.granularity=['low','high']`. Defaults to `auto`.
+* `dtw.granularity`: adjusts the MFCC frame width and hop size based on the profile selected. Can be set to either `xx-low` (400ms width, 160ms hop), `x-low` (200ms width, 80ms hop), `low` (100ms width, 40ms hop), `medium` (50ms width, 20ms hop), `high` (25ms width, 10ms hop), `x-high` (20ms width, 5ms hop). For multi-pass processing, multiple granularities can be provided, like `dtw.granularity=['xx-low','medium']`. Auto-selected by default.
 * `dtw.windowDuration`: maximum duration (in seconds) of the Sakoe-Chiba window when performing DTW alignment. Higher values consume quadratically larger amounts of memory. The estimated memory requirement is shown in the log before alignment starts. Recommended to be set to at least 10% - 20% of total audio duration. For multi-pass processing, multiple durations can be provided, like `dtw.windowDuration=[240,20]`. Auto-selected by default
 
 **DTW-RA**:

package/docs/Tasklist.md

@@ -9,7 +9,8 @@
 
 ### Synthesis
 
-### Phoneme processing
+### eSpeak
+
 * IPA -> Kirshenbaum translation is still not completely similar to what is output by eSpeak. Also, in rare situations, it outputs characters that are not accepted by eSpeak and eSpeak errors. Investigate when that happens and how to improve on this
 
 ### Browser extension
@@ -151,8 +152,7 @@
 * Show alternatives when playing in the CLI. Clear current line and rewrite already printed text for alternatives during the speech recognition process
 
 ### Recognition / Whisper
-* Whisper's Chinese and Japanese output can be split into words in a more accurate way. Consider using a dedicated segmentation library to perform the segmentation in character sequences that have no spaces within them
-* Automatically disable using previous section recognized transcript as prompt for the next section when lots of repetition occurred in previous section
+* Whisper's Chinese and Japanese output can be split into words in a more accurate way. Consider using a dedicated segmentation library to perform the segmentation in character sequences that have no punctuation characters to aid on guessing word boundaries
 * Cache last model (if enough memory is available)
 * The segment output can be used to split into segments, otherwise it is possible to try to guess using pause lengths or voice activity detection
 * Bring back the option to use eSpeak DTW based alignment on segments, as an alternative approach

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "echogarden",
-	"version": "1.3.2",
+	"version": "1.4.0",
 	"description": "An easy-to-use speech toolset. Includes tools for synthesis, recognition, alignment, speech translation, language detection, source separation and more.",
 	"author": "Rotem Dan",
 	"license": "GPL-3.0",
@@ -55,8 +55,8 @@
 		"echogarden": "./dist/cli/CLILauncher.js"
 	},
 	"dependencies": {
-		"@aws-sdk/client-polly": "^3.569.0",
-		"@aws-sdk/client-transcribe-streaming": "^3.569.0",
+		"@aws-sdk/client-polly": "^3.574.0",
+		"@aws-sdk/client-transcribe-streaming": "^3.574.0",
 		"@echogarden/espeak-ng-emscripten": "^0.1.2",
 		"@echogarden/fasttext-wasm": "^0.1.0",
 		"@echogarden/flite-wasi": "^0.1.1",
@@ -89,14 +89,14 @@
 		"moving-median": "^1.0.0",
 		"msgpack-lite": "^0.1.26",
 		"onnxruntime-node": "^1.17.3",
-		"openai": "^4.42.0",
+		"openai": "^4.45.0",
 		"sam-js": "^0.2.1",
 		"strip-ansi": "^7.1.0",
 		"tar": "^7.1.0",
 		"tiktoken": "^1.0.14",
 		"tinyld": "^1.3.4",
 		"ws": "^8.17.0",
-		"wtf_wikipedia": "^10.3.0"
+		"wtf_wikipedia": "^10.3.1"
 	},
 	"peerDependencies": {
 		"@echogarden/vosk": "^0.3.39-patched.1",
@@ -120,11 +120,11 @@
 		"@types/graceful-fs": "^4.1.9",
 		"@types/jsdom": "^21.1.6",
 		"@types/msgpack-lite": "^0.1.11",
-		"@types/node": "^20.12.10",
+		"@types/node": "^20.12.11",
 		"@types/recursive-readdir": "^2.2.4",
 		"@types/tar": "^6.1.13",
 		"@types/ws": "^8.5.10",
-		"ts-json-schema-generator": "^2.1.1",
+		"ts-json-schema-generator": "^2.1.2-next.1",
 		"typescript": "^5.4.5"
 	}
 }

package/src/alignment/DTWSequenceAlignmentWindowed.ts

@@ -1,5 +1,3 @@
-import chalk from 'chalk'
-import { Logger } from '../utilities/Logger.js'
 import { logToStderr } from '../utilities/Utilities.js'
 import { AlignmentPath } from './SpeechAlignment.js'
 
@@ -11,7 +9,10 @@ export function alignDTWWindowed<T, U>(sequence1: T[], sequence2: U[], costFunct
 	}
 
 	if (sequence1.length == 0 || sequence2.length == 0) {
-		return { path: [] as AlignmentPath, pathCost: 0 }
+		return {
+			path: [] as AlignmentPath,
+			pathCost: 0
+		}
 	}
 
 	// Compute accumulated cost matrix (transposed)
@@ -38,10 +39,10 @@ function computeAccumulatedCostMatrixTransposed<T, U>(sequence1: T[], sequence2:
 
 	const accumulatedCostMatrixTransposed: Float32Array[] = new Array<Float32Array>(columnCount)
 
-	// Initialize window start offsets array
+	// Initialize an array to store window start offsets
 	const windowStartOffsets = new Int32Array(columnCount)
 
-	// Compute matrix column by column
+	// Compute accumulated cost matrix column by column
 	for (let columnIndex = 0; columnIndex < columnCount; columnIndex++) {
 		// Create new column and add it to the matrix
 		const currentColumn = new Float32Array(rowCount)
@@ -65,13 +66,13 @@ function computeAccumulatedCostMatrixTransposed<T, U>(sequence1: T[], sequence2:
 			windowStartOffset = windowEndOffset - rowCount
 		}
 
-		// Store the start offset in the array
+		// Store the start offset for this column
 		windowStartOffsets[columnIndex] = windowStartOffset
 
 		// Get target sequence1 value
 		const targetSequence1Value = sequence1[columnIndex]
 
-		// If first column, fill it only using the 'up' neighbor
+		// If this is the first column, fill it only using the 'up' neighbors
 		if (columnIndex == 0) {
 			for (let rowIndex = 1; rowIndex < rowCount; rowIndex++) {
 				const cost = costFunction(targetSequence1Value, sequence2[windowStartOffset + rowIndex])
@@ -84,11 +85,15 @@ function computeAccumulatedCostMatrixTransposed<T, U>(sequence1: T[], sequence2:
 		}
 
 		// If not first column
+
+		// Store the column to the left
 		const leftColumn = accumulatedCostMatrixTransposed[columnIndex - 1]
 
-		// Compute the delta between the current window offset and previous column's window offset
+		// Compute the delta between the current window start offset
+		// and left column's window offset
 		const windowOffsetDelta = windowStartOffset - windowStartOffsets[columnIndex - 1]
 
+		// Iterate over all rows in the window
 		for (let rowIndex = 0; rowIndex < rowCount; rowIndex++) {
 			// Compute the cost for current cell
 			const cost = costFunction(targetSequence1Value, sequence2[windowStartOffset + rowIndex])
@@ -115,12 +120,25 @@ function computeAccumulatedCostMatrixTransposed<T, U>(sequence1: T[], sequence2:
 				upAndLeftCost = leftColumn[upAndLeftRowIndex]
 			}
 
+			// Find the minimum of all neighbors
+			let minimumNeighborCost = minimumOf3(upCost, leftCost, upAndLeftCost)
+
+			// If all neighbors are infinity, then it means there is a "jump" between the window
+			// of the current column and the left column, and they don't have overlapping rows.
+			// In this case, only the cost of the current cell will be used
+			if (minimumNeighborCost === Infinity) {
+				minimumNeighborCost = 0
+			}
+
 			// Write cost + minimum neighbor cost to the current column
-			currentColumn[rowIndex] = cost + minimumOf3(upCost, leftCost, upAndLeftCost)
+			currentColumn[rowIndex] = cost + minimumNeighborCost
 		}
 	}
 
-	return { accumulatedCostMatrixTransposed, windowStartOffsets }
+	return {
+		accumulatedCostMatrixTransposed,
+		windowStartOffsets
+	}
 }
 
 function computeBestPathTransposed(accumulatedCostMatrixTransposed: Float32Array[], windowStartOffsets: Int32Array) {
@@ -129,6 +147,8 @@ function computeBestPathTransposed(accumulatedCostMatrixTransposed: Float32Array
 
 	const bestPath: AlignmentPath = []
 
+	// Start at the bottom right corner and find the best path
+	// towards the top left
 	let columnIndex = columnCount - 1
 	let rowIndex = rowCount - 1
 
@@ -136,19 +156,21 @@ function computeBestPathTransposed(accumulatedCostMatrixTransposed: Float32Array
 		const windowStartIndex = windowStartOffsets[columnIndex]
 		const windowStartDelta = columnIndex > 0 ? windowStartIndex - windowStartOffsets[columnIndex - 1] : 0
 
+		// Add the current cell to the best path
 		bestPath.push({
 			source: columnIndex,
 			dest: windowStartIndex + rowIndex
 		})
 
+		// Retrieve the cost for the 'up' (insertion) neighbor
 		const upRowIndex = rowIndex - 1
-		const upColumnIndex = columnIndex
 		let upCost = Infinity
 
 		if (upRowIndex >= 0) {
-			upCost = accumulatedCostMatrixTransposed[upColumnIndex][upRowIndex] // insertion
+			upCost = accumulatedCostMatrixTransposed[columnIndex][upRowIndex] // insertion
 		}
 
+		// Retrieve the cost for the 'left' (deletion) neighbor
 		const leftRowIndex = rowIndex + windowStartDelta
 		const leftColumnIndex = columnIndex - 1
 		let leftCost = Infinity
@@ -157,6 +179,7 @@ function computeBestPathTransposed(accumulatedCostMatrixTransposed: Float32Array
 			leftCost = accumulatedCostMatrixTransposed[leftColumnIndex][leftRowIndex] // deletion
 		}
 
+		// Retrieve the cost for the 'up and left' (match) neighbor
 		const upAndLeftRowIndex = rowIndex - 1 + windowStartDelta
 		const upAndLeftColumnIndex = columnIndex - 1
 		let upAndLeftCost = Infinity
@@ -165,25 +188,42 @@ function computeBestPathTransposed(accumulatedCostMatrixTransposed: Float32Array
 			upAndLeftCost = accumulatedCostMatrixTransposed[upAndLeftColumnIndex][upAndLeftRowIndex] // match
 		}
 
+		// If all neighbors have a cost of infinity, it means
+		// there is a "jump" between the window for the current and previous column
 		if (upCost == Infinity && leftCost == Infinity && upAndLeftCost == Infinity) {
-			const logger = new Logger()
-
-			logger.setAsActiveLogger()
-			logger.logTitledMessage(`computeBestPath`, `unexpected - all cost directions are equal to infinity (${columnIndex}, ${rowIndex}).`, chalk.yellowBright, 'warning')
-			logger.unsetAsActiveLogger()
-		}
-
-		const smallestCostDirection = argIndexOfMinimumOf3(upCost, leftCost, upAndLeftCost)
-
-		if (smallestCostDirection == 1) {
-			rowIndex = upRowIndex
-			columnIndex = upColumnIndex
-		} else if (smallestCostDirection == 2) {
-			rowIndex = leftRowIndex
-			columnIndex = leftColumnIndex
+			// In that case:
+			//
+			// If there are rows above
+			if (upRowIndex >= 0) {
+				// Move upward
+				rowIndex = upRowIndex
+			} else if (leftColumnIndex >= 0) {
+				// Otherwise, move to the left
+				columnIndex = leftColumnIndex
+			} else {
+				// Since we know that either columnIndex > 0 or rowIndex > 0,
+				// one of these directions must be available.
+				// This error should never happen
+
+				throw new Error(`Unexpected state: columnIndex: ${columnIndex}, rowIndex: ${rowIndex}`)
+			}
 		} else {
-			rowIndex = upAndLeftRowIndex
-			columnIndex = upAndLeftColumnIndex
+			// Choose the direction with the smallest cost
+			const smallestCostDirection = argIndexOfMinimumOf3(upCost, leftCost, upAndLeftCost)
+
+			if (smallestCostDirection == 1) {
+				// Move upward
+				rowIndex = upRowIndex
+				// The upper column index stays the same
+			} else if (smallestCostDirection == 2) {
+				// Move to the left
+				rowIndex = leftRowIndex
+				columnIndex = leftColumnIndex
+			} else {
+				// Move upward and to the left
+				rowIndex = upAndLeftRowIndex
+				columnIndex = upAndLeftColumnIndex
+			}
 		}
 	}
 

package/src/alignment/SpeechAlignment.ts

@@ -38,12 +38,12 @@ export async function alignUsingDtw(
 	let relativeCenters: number[] | undefined
 
 	for (let passIndex = 0; passIndex < windowDurations.length; passIndex++) {
+		const granularity = granularities[passIndex]
 		const windowDuration = windowDurations[passIndex]
-		const granularity = resolveAutoGranularityIfNeeded(granularities[passIndex], rawAudioDuration)
 
 		logger.logTitledMessage(`\nStarting alignment pass ${passIndex + 1}/${windowDurations.length}`, `max window duration: ${windowDuration}s, granularity: ${granularity}`, chalk.magentaBright)
 
-		const mfccOptions = extendDefaultMfccOptions({ ...getMfccOptionsForGranularity(granularity, rawAudioDuration), zeroFirstCoefficient: true }) as MfccOptions
+		const mfccOptions = extendDefaultMfccOptions({ ...getMfccOptionsForGranularity(granularity), zeroFirstCoefficient: true }) as MfccOptions
 
 		framesPerSecond = 1 / mfccOptions.hopDuration!
 
@@ -67,7 +67,7 @@ export async function alignUsingDtw(
 			}
 		}
 
-		logger.start('Align MFCC features using DTW')
+		logger.start('Align reference and source MFCC features using DTW')
 		const dtwWindowLength = Math.floor(windowDuration * framesPerSecond)
 
 		let centerIndexes: number[] | undefined
@@ -458,7 +458,7 @@ export async function createAlignmentReferenceUsingEspeakForFragments(fragments:
 	progressLogger.start("Load espeak module")
 	const Espeak = await import("../synthesis/EspeakTTS.js")
 
-	progressLogger.start("Create alignment reference with eSpeak")
+	progressLogger.start("Synthesize alignment reference with eSpeak")
 
 	const result = await Espeak.synthesizeFragments(fragments, espeakOptions)
 
@@ -476,7 +476,7 @@ export async function createAlignmentReferenceUsingEspeakForFragments(fragments:
 export async function createAlignmentReferenceUsingEspeak(transcript: string, language: string, plaintextOptions?: API.PlainTextOptions, customLexiconPaths?: string[], insertSeparators?: boolean) {
 	const logger = new Logger()
 
-	logger.start('Create alignment reference with eSpeak')
+	logger.start('Synthesize alignment reference with eSpeak')
 
 	const synthesisOptions: API.SynthesisOptions = {
 		engine: 'espeak',
@@ -544,25 +544,9 @@ function getMappedFrameIndexForPath(referenceFrameIndex: number, compactedPath:
 	return mappedFrameIndex
 }
 
-function resolveAutoGranularityIfNeeded(granularity: DtwGranularity, audioDuration: number) {
-	if (granularity != 'auto') {
-		return granularity
-	}
-
-	if (audioDuration < 60) {
-		return 'high'
-	} else if (audioDuration < 60 * 10) {
-		return 'medium'
-	} else {
-		return 'low'
-	}
-}
-
-function getMfccOptionsForGranularity(granularity: DtwGranularity, audioDuration: number) {
+function getMfccOptionsForGranularity(granularity: DtwGranularity) {
 	let mfccOptions: MfccOptions
 
-	granularity = resolveAutoGranularityIfNeeded(granularity, audioDuration)
-
 	if (granularity == 'xx-low') {
 		mfccOptions = { windowDuration: 0.400, hopDuration: 0.160, fftOrder: 8192 }
 	} else if (granularity == 'x-low') {
@@ -595,4 +579,4 @@ export type CompactedPathEntry = {
 	first: number, last: number
 }
 
-export type DtwGranularity = 'auto' | 'xx-low' | 'x-low' | 'low' | 'medium' | 'high' | 'x-high'
+export type DtwGranularity = 'xx-low' | 'x-low' | 'low' | 'medium' | 'high' | 'x-high'